You are a constitutional council ranking individual git commits for ownership allocation.

Compare these two commits. Decide which contributed more lasting value to the project.

Judge substance, not spectacle:
- Prefer correct, lasting design and real bugfixes over churn, formatting, renames, or generated noise.
- Prefer clarity and necessity over sheer line count. A small precise change can beat a large diffuse one.
- Do not favor a side merely because its patch is longer or noisier.
- Weight what the change does for the project, not the contributor's name.

Return ONLY a JSON object: {"winner": "A" or "B", "ratio": "N:M", "explanation": "..."}
The explanation must cite concrete differences in the patches (1-3 sentences).

Side A — contributor: tommy-mor
Side A — commit message:
[a888d56c] refactor: centralize path identity in slug-types

Move canonicalization and CanonicalItemUrl into types::paths with
GardenItemUrl, ForumThreadUrl, and TildeOntologyPath for JSON hrefs.
Server canonical_path and path_types re-export slug-types; RPC and
validation build hrefs via those types instead of string helpers.

Made-with: Cursor

Side A — unified diff (full patch):
diff --git a/server/src/api/helpers.rs b/server/src/api/helpers.rs
index 9b71491e9f9efc44a2a4beba09be8f64bd2ff2ee..03b3e77911ccd662bec8635345dafe2593cf242e 100644
--- a/server/src/api/helpers.rs
+++ b/server/src/api/helpers.rs
@@ -4,12 +4,12 @@ use axum::{
     Json,
 };
 use sha2::{Digest, Sha256};
+use slug_types::paths::{CanonicalItemUrl, GardenItemUrl};
 use slug_types::*;
 use std::collections::HashMap;
 
 use crate::{
     canonical_path::canonicalize_item,
-    path_types::CanonicalItemUrl,
     ranking::connected_components_from_voted_pairs,
 };
 
@@ -30,64 +30,6 @@ pub fn now_ms() -> i64 {
     t.as_millis() as i64
 }
 
-/// Serialize a canonical item for JSON: absolute URLs stay as-is; bare paths get a `/` prefix.
-pub fn item_path_for_api(item: &str) -> String {
-    if item.starts_with("http://") || item.starts_with("https://") {
-        item.to_string()
-    } else {
-        format!("/{}", item)
-    }
-}
-
-/// Same as [`item_path_for_api`], but for private rooms ontology items are prefixed with
-/// `/r/{short}/{slug}` so the URL matches the web app (`/r/…/~/…` routes).
-pub fn item_path_for_api_in_room(item: &str, room_wire: &str) -> String {
-    let room = room_wire.trim();
-    if room.is_empty() || room == "public" {
-        return item_path_for_api(item);
-    }
-    let Some((short, slug)) = room.split_once('/') else {
-        return item_path_for_api(item);
-    };
-    if short.is_empty() || slug.is_empty() {
-        return item_path_for_api(item);
-    }
-    let Some(c) = CanonicalItemUrl::parse(item) else {
-        return item_path_for_api(item);
-    };
-    let root = CanonicalItemUrl::ontology_root();
-    let item_norm = c.as_str().trim_end_matches('/');
-    let root_norm = root.as_str().trim_end_matches('/');
-    if let Some(tail) = c.tilde_tail() {
-        return if tail.is_empty() {
-            format!("https://slug.social/r/{short}/{slug}/~")
-        } else {
-            format!("https://slug.social/r/{short}/{slug}/~/{}", tail)
-        };
-    }
-    if item_norm == root_norm {
-        return format!("https://slug.social/r/{short}/{slug}/~");
-    }
-    item_path_for_api(item)
-}
-
-/// Absolute thread URL for forum JSON (`/t/…` vs `/r/…/t/…`).
-pub fn forum_thread_web_url(room_wire: &str, thread_tag: &str) -> String {
-    let room = room_wire.trim();
-    let tag = thread_tag.trim().trim_start_matches('#');
-    if room.is_empty() || room == "public" {
-        format!("https://slug.social/t/{tag}")
-    } else if let Some((short, slug)) = room.split_once('/') {
-        if short.is_empty() || slug.is_empty() {
-            format!("https://slug.social/t/{tag}")
-        } else {
-            format!("https://slug.social/r/{short}/{slug}/t/{tag}")
-        }
-    } else {
-        format!("https://slug.social/t/{tag}")
-    }
-}
-
 /// Resolve an item path as a first-class canonical path.
 pub fn resolve_item(item: &str) -> Result<String, String> {
     let canonical = canonicalize_item(item);
@@ -109,14 +51,12 @@ pub fn parse_parent_specs(parent: Option<&String>) -> Vec<String> {
 }
 
 /// Apply offset+limit pagination to the flattened component rankings.
-/// Items are flattened in component order (largest component first), then unranked last.
-/// Returns (components, unranked_items) after the window.
 pub fn paginate_rankings(
     components: Vec<RankComponent>,
-    unranked_items: Vec<String>,
+    unranked_items: Vec<GardenItemUrl>,
     offset: usize,
     limit: Option<usize>,
-) -> (Vec<RankComponent>, Vec<String>) {
+) -> (Vec<RankComponent>, Vec<GardenItemUrl>) {
     let mut remaining_skip = offset;
     let mut remaining_take = limit.unwrap_or(usize::MAX);
     let mut out_components: Vec<RankComponent> = Vec::new();
@@ -141,7 +81,7 @@ pub fn paginate_rankings(
         });
     }
 
-    let out_unranked: Vec<String> = if remaining_take > 0 {
+    let out_unranked: Vec<GardenItemUrl> = if remaining_take > 0 {
         unranked_items
             .into_iter()
             .skip(remaining_skip)
@@ -183,11 +123,9 @@ pub fn is_pair_voted(group: &crate::reducer::GroupState, a: &str, b: &str) -> bo
     group.voted_pairs.contains(&(i, j))
 }
 
-/// Compute graph connectivity stats for a set of items within the ranking group.
 pub fn compute_connectivity_stats(group: &crate::reducer::GroupState, pool: &[String]) -> ConnectivityStats {
     let n = pool.len();
 
-    // Map pool items to global indices (items not yet in the group get no index)
     let global_idxs: Vec<Option<usize>> = pool
         .iter()
         .map(|it| {
@@ -197,7 +135,6 @@ pub fn compute_connectivity_stats(group: &crate::reducer::GroupState, pool: &[St
         .collect();
     let present: Vec<usize> = global_idxs.iter().filter_map(|x| *x).collect();
 
-    // Build local index mapping for items that exist in the ranking group
     let global_to_local: HashMap<usize, usize> = present
         .iter()
         .enumerate()
@@ -213,7 +150,6 @@ pub fn compute_connectivity_stats(group: &crate::reducer::GroupState, pool: &[St
         }),
     );
 
-    // Items not in the ranking group at all are also isolates
     let items_not_in_group = global_idxs.iter().filter(|x| x.is_none()).count();
 
     let num_components = comps.len() + isolates.len() + items_not_in_group;
@@ -237,52 +173,3 @@ pub fn vote_touches_path(a: &str, b: &str, parent_canon: &str) -> bool {
     let under = |item: &str| item == parent_canon || item.starts_with(&format!("{}/", parent_canon));
     under(a) || under(b)
 }
-
-#[cfg(test)]
-mod wire_url_tests {
-    use super::{forum_thread_web_url, item_path_for_api_in_room};
-
-    #[test]
-    fn public_room_unchanged() {
-        let u = "https://slug.social/~/a/b";
-        assert_eq!(item_path_for_api_in_room(u, "public"), u);
-    }
-
-    #[test]
-    fn private_room_prefixes_ontology() {
-        assert_eq!(
-            item_path_for_api_in_room("https://slug.social/~/topic/x", "9ab12cd/my-room"),
-            "https://slug.social/r/9ab12cd/my-room/~/topic/x"
-        );
-    }
-
-    #[test]
-    fn private_room_ontology_root() {
-        assert_eq!(
-            item_path_for_api_in_room("https://slug.social/~", "9ab12cd/my-room"),
-            "https://slug.social/r/9ab12cd/my-room/~"
-        );
-        assert_eq!(
-            item_path_for_api_in_room("https://slug.social/~/", "9ab12cd/my-room"),
-            "https://slug.social/r/9ab12cd/my-room/~"
-        );
-    }
-
-    #[test]
-    fn external_url_untouched_in_private_room() {
-        let u = "https://example.com/z";
-        assert_eq!(item_path_for_api_in_room(u, "9ab12cd/my-room"), u);
-    }
-
-    #[test]
-    fn forum_web_public_vs_room() {
-        assert_eq!(
-            forum_thread_web_url("public", "debate"),
-            "https://slug.social/t/debate"
-        );
-        assert_eq!(
-            forum_thread_web_url("9ab12cd/my-room", "#debate"),
-            "https://slug.social/r/9ab12cd/my-room/t/debate"
-        );
-    }
-}
diff --git a/server/src/api/mod.rs b/server/src/api/mod.rs
index 042aa248305f9362a3be78f9eea2a5abf6ba707a..cf22cb0129366c3aed031bc86f3197a4321cb806 100644
--- a/server/src/api/mod.rs
+++ b/server/src/api/mod.rs
@@ -24,8 +24,7 @@ pub use auth::{
 
 pub use helpers::{
     api_error, compute_connectivity_stats, is_pair_voted, now_ms, paginate_rankings,
-    parse_parent_specs, pick_random_distinct, sha256_hex, resolve_item, vote_touches_path,
-    item_path_for_api,
+    parse_parent_specs, pick_random_distinct, resolve_item, sha256_hex, vote_touches_path,
 };
 
 pub use rpc::handle_rpc_batch;
diff --git a/server/src/api/rpc.rs b/server/src/api/rpc.rs
index 5b91f5836625eedbb1cd9423168046e3fb576c17..5f7d50188f1381267402f2e57e671234ef5db2fd 100644
--- a/server/src/api/rpc.rs
+++ b/server/src/api/rpc.rs
@@ -8,6 +8,7 @@ use axum::{
     Json,
 };
 use rand::seq::SliceRandom;
+use slug_types::paths::{ForumThreadUrl, GardenItemUrl, TildeOntologyPath};
 use slug_types::*;
 
 use crate::{
@@ -27,9 +28,8 @@ use crate::{
 
 use super::auth::verify_bearer_principal;
 use super::helpers::{
-    compute_connectivity_stats, forum_thread_web_url, is_pair_voted, item_path_for_api,
-    item_path_for_api_in_room, now_ms, paginate_rankings, parse_parent_specs, pick_random_distinct,
-    resolve_item, vote_touches_path,
+    compute_connectivity_stats, is_pair_voted, now_ms, paginate_rankings, parse_parent_specs,
+    pick_random_distinct, resolve_item, vote_touches_path,
 };
 use super::validate::{normalize_room_and_thread, validate_ingest_document};
 
@@ -184,7 +184,7 @@ fn compute_scope_rank_changes(
         };
         if changed {
             changes.push(RankChange {
-                item: item_path_for_api_in_room(&item, room_wire),
+                item: GardenItemUrl::from_storage_str(&item, room_wire),
                 before: b,
                 after: a,
             });
@@ -206,7 +206,7 @@ fn compute_scope_rank_changes(
         parent: if parent.is_empty() {
             "/".to_string()
         } else {
-            item_path_for_api_in_room(parent, room_wire)
+            GardenItemUrl::from_storage_str(parent, room_wire).into_inner()
         },
         changes,
     })
@@ -302,7 +302,7 @@ fn build_rank_response_for_content(
                     .ranked
                     .into_iter()
                     .map(|r| RankRow {
-                        item: item_path_for_api_in_room(r.item.as_str(), room_wire),
+                        item: GardenItemUrl::from_stored(&r.item, room_wire),
                         percent: if want_percent {
                             Some((r.score / max_score) * 100.0)
                         } else {
@@ -315,10 +315,10 @@ fn build_rank_response_for_content(
         })
         .collect();
 
-    let prefixed_unranked: Vec<String> = rankings
+    let prefixed_unranked: Vec<GardenItemUrl> = rankings
         .unranked_items
         .into_iter()
-        .map(|s| item_path_for_api_in_room(s.as_str(), room_wire))
+        .map(|s| GardenItemUrl::from_stored(&s, room_wire))
         .collect();
 
     let (components, unranked_items) = if offset > 0 || limit.is_some() {
@@ -537,13 +537,13 @@ async fn rpc_post(
         (
             "npx slugsocial public garden pair".to_string(),
             "npx slugsocial public garden rank".to_string(),
-            forum_thread_web_url("public", &thread_id),
+            ForumThreadUrl::from_room_tag("public", &thread_id),
         )
     } else {
         (
             format!("npx slugsocial private {room_key} garden pair"),
             format!("npx slugsocial private {room_key} garden rank"),
-            forum_thread_web_url(&room_key, &thread_id),
+            ForumThreadUrl::from_room_tag(&room_key, &thread_id),
         )
     };
 
@@ -664,7 +664,7 @@ async fn rpc_check(
                         .ranked
                         .into_iter()
                         .map(|r| RankRow {
-                            item: item_path_for_api_in_room(r.item.as_str(), &room_key),
+                            item: GardenItemUrl::from_stored(&r.item, &room_key),
                             score: r.score,
                             percent: None,
                         })
@@ -672,12 +672,12 @@ async fn rpc_check(
                 })
                 .collect();
             CheckScopeRanking {
-                parent: item_path_for_api_in_room(parent.as_str(), &room_key),
+                parent: GardenItemUrl::from_stored(parent, &room_key).into_inner(),
                 components,
                 unranked_items: scoped
                     .unranked_items
                     .into_iter()
-                    .map(|it| item_path_for_api_in_room(it.as_str(), &room_key))
+                    .map(|it| GardenItemUrl::from_stored(&it, &room_key))
                     .collect(),
             }
         })
@@ -687,13 +687,13 @@ async fn rpc_check(
         vec![
             "npx slugsocial public forum post <TAG> --delegate <uuid:rig:model>".to_string(),
             "npx slugsocial public forum list".to_string(),
-            forum_thread_web_url("public", &thread_id),
+            ForumThreadUrl::from_room_tag("public", &thread_id).into_inner(),
         ]
     } else {
         vec![
             format!("npx slugsocial private {room_key} forum post <TAG> --delegate <uuid:rig:model>"),
             format!("npx slugsocial private {room_key} forum list"),
-            forum_thread_web_url(&room_key, &thread_id),
+            ForumThreadUrl::from_room_tag(&room_key, &thread_id).into_inner(),
         ]
     };
 
@@ -717,7 +717,7 @@ fn rpc_list_forum_threads(reduced: &ReducerState, room: &str) -> ThreadsResponse
         .map(|((_, tag), ts)| ThreadSummary {
             thread: format!("#{tag}"),
             last_activity_ts: ts.last_activity_ts,
-            web: forum_thread_web_url(room, tag),
+            web: ForumThreadUrl::from_room_tag(room, tag),
         })
         .collect();
     out.sort_by(|a, b| b.last_activity_ts.cmp(&a.last_activity_ts));
@@ -885,7 +885,7 @@ fn rpc_search(reduced: &ReducerState, q: &str, limit: usize, principal: Option<&
         }
         if score > 0 {
             scored_items.push((score, SearchItemHit {
-                path: item_path_for_api(item.as_str()),
+                path: GardenItemUrl::from_storage_str(item.as_str(), "public"),
                 body: content.item_bodies.get(item).map(|b| snippet_around(b, &words, 120)),
             }));
         }
@@ -1058,8 +1058,8 @@ async fn rpc_get_pair(state: &AppState, room: String, parent_path: String) -> Re
         .collect();
     let cs = compute_connectivity_stats(&content.ranking_group, &pool);
     Ok(RpcResult::Pair(PairResponse {
-        left: item_path_for_api_in_room(&left, &room),
-        right: item_path_for_api_in_room(&right, &room),
+        left: GardenItemUrl::from_storage_str(&left, &room),
+        right: GardenItemUrl::from_storage_str(&right, &room),
         left_body: lb,
         right_body: rb,
         threads: th,
@@ -1137,7 +1137,7 @@ pub async fn handle_rpc_batch(
                 if !content.items.contains(&item) {
                     line_err(
                         "item not found",
-                        Some(format!("{} does not exist", item_path_for_api_in_room(&item_str, &room))),
+                        Some(format!("{} does not exist", GardenItemUrl::from_storage_str(&item_str, &room))),
                     )
                 } else {
                     const MAX_ITEM_BODY: usize = 10_000;
@@ -1160,7 +1160,7 @@ pub async fn handle_rpc_batch(
                         .map(|s| s.iter().cloned().collect())
                         .unwrap_or_default();
                     line_ok(RpcResult::GardenItem(ItemResponse {
-                        item: item_path_for_api_in_room(&item_str, &room),
+                        item: GardenItemUrl::from_storage_str(&item_str, &room),
                         body,
                         truncated,
                         body_len,
@@ -1554,7 +1554,7 @@ pub async fn handle_rpc_batch(
                     for r in items {
                         let pct = want_percent.then(|| ((r.score - bot) / range * 100.0).clamp(0.0, 100.0));
                         ranked.push(RankRow {
-                            item: item_path_for_api_in_room(r.item.as_str(), &room),
+                            item: GardenItemUrl::from_storage_str(r.item.as_str(), &room),
                             score: r.score,
                             percent: pct,
                         });
@@ -1574,7 +1574,7 @@ pub async fn handle_rpc_batch(
                 let page: Vec<RankRow> = ranked
                     .into_iter()
                     .chain(unranked.into_iter().map(|it| RankRow {
-                        item: item_path_for_api_in_room(&it, &room),
+                        item: GardenItemUrl::from_storage_str(&it, &room),
                         score: 0.0,
                         percent: want_percent.then_some(0.0),
                     }))
@@ -1618,7 +1618,7 @@ pub async fn handle_rpc_batch(
                 if !content.items.contains(&item) {
                     line_err(
                         "item not found",
-                        Some(format!("{} does not exist", item_path_for_api_in_room(&item_str, &room))),
+                        Some(format!("{} does not exist", GardenItemUrl::from_storage_str(&item_str, &room))),
                     )
                 } else {
                     let votes: Vec<VoteRow> = content
@@ -1629,8 +1629,8 @@ pub async fn handle_rpc_batch(
                                 .take(limit)
                                 .map(|v| VoteRow {
                                     ts: v.ts,
-                                    a: item_path_for_api_in_room(v.a.as_str(), &room),
-                                    b: item_path_for_api_in_room(v.b.as_str(), &room),
+                                    a: GardenItemUrl::from_stored(&v.a, &room),
+                                    b: GardenItemUrl::from_stored(&v.b, &room),
                                     ratio: format!("{}:{}", v.ratio_left, v.ratio_right),
                                     actor: Some(v.principal.clone()),
                                     body: v.body.clone(),
@@ -1640,7 +1640,7 @@ pub async fn handle_rpc_batch(
                         })
                         .unwrap_or_default();
                     line_ok(RpcResult::Matchup(MatchupResponse {
-                        item: item_path_for_api_in_room(&item_str, &room),
+                        item: GardenItemUrl::from_storage_str(&item_str, &room),
                         votes,
                     }))
                 }
@@ -1667,8 +1667,8 @@ pub async fn handle_rpc_batch(
                                     if a == item_str || b == item_str {
                                         Some(VoteRow {
                                             ts: e.ts,
-                                            a: item_path_for_api_in_room(&a, &room),
-                                            b: item_path_for_api_in_room(&b, &room),
+                                            a: GardenItemUrl::from_storage_str(&a, &room),
+                                            b: GardenItemUrl::from_storage_str(&b, &room),
                                             ratio: format!("{}:{}", ratio_left, ratio_right),
                                             actor: reduced.ingests_by_id.get(&e.post_id).map(|ing| ing.principal.clone()),
                                             body: explanation,
@@ -1704,7 +1704,7 @@ pub async fn handle_rpc_batch(
                     }
                 }).collect();
                 line_ok(RpcResult::RankHistory(RankHistoryResponse {
-                    item: item_path_for_api_in_room(&item_str, &room),
+                    item: GardenItemUrl::from_storage_str(&item_str, &room),
                     history,
                 }))
                 }
@@ -1716,17 +1716,13 @@ pub async fn handle_rpc_batch(
                 } else {
                 let content = content_for_room(&reduced, &room);
                 let parents: HashSet<&str> = content.item_children.keys().map(|s| s.as_str()).collect();
-                let mut paths: Vec<String> = content
+                let mut paths: Vec<GardenItemUrl> = content
                     .items
                     .iter()
                     .filter(|p| !parents.contains(p.as_str()))
-                    .map(|p| p.as_str().to_string())
-                    .collect();
-                paths.sort();
-                let paths: Vec<String> = paths
-                    .into_iter()
-                    .map(|p| item_path_for_api_in_room(&p, &room))
+                    .map(|p| GardenItemUrl::from_stored(p, &room))
                     .collect();
+                paths.sort_by(|a, b| a.as_str().cmp(b.as_str()));
                 line_ok(RpcResult::Leaves(LeavesResponse { paths }))
                 }
             },
@@ -1743,24 +1739,13 @@ pub async fn handle_rpc_batch(
                         let mut v: Vec<PathSummary> = roots.iter()
                             .map(|path| {
                                 let children = content.item_children.get(path.as_str()).map(|s| s.len()).unwrap_or(0);
-                                let path_label = CanonicalItemUrl::parse(path.as_str())
-                                    .and_then(|c| {
-                                        c.tilde_tail().map(|t| {
-                                            if t.is_empty() {
-                                                "~/".to_string()
-                                            } else {
-                                                format!("~/{}", t)
-                                            }
-                                        })
-                                    })
-                                    .unwrap_or_else(|| path.to_string());
                                 PathSummary {
-                                    path: path_label,
+                                    path: TildeOntologyPath::from_stored(path),
                                     children,
-                                    web: item_path_for_api_in_room(path.as_str(), &room),
+                                    web: GardenItemUrl::from_stored(path, &room),
                                 }
                             }).collect();
-                        v.sort_by(|a, b| a.path.cmp(&b.path));
+                        v.sort_by(|a, b| a.path.as_str().cmp(b.path.as_str()));
                         v
                     })
                     .unwrap_or_default();
@@ -1790,8 +1775,8 @@ pub async fn handle_rpc_batch(
                     .take(limit)
                     .map(|v| VoteRow {
                         ts: v.ts,
-                        a: item_path_for_api_in_room(v.a.as_str(), &room),
-                        b: item_path_for_api_in_room(v.b.as_str(), &room),
+                        a: GardenItemUrl::from_stored(&v.a, &room),
+                        b: GardenItemUrl::from_stored(&v.b, &room),
                         ratio: format!("{}:{}", v.ratio_left, v.ratio_right),
                         actor: Some(v.principal.clone()),
                         body: v.body.clone(),
diff --git a/server/src/api/validate.rs b/server/src/api/validate.rs
index 27150577982f52cbd6d11bb93654dc3d4cf75cc5..a51c783ee9785569b5a44c0b1572471fe00d174b 100644
--- a/server/src/api/validate.rs
+++ b/server/src/api/validate.rs
@@ -7,8 +7,9 @@ use crate::{
     path_types::CanonicalItemUrl,
     reducer::{ReducerState, ScopeId},
 };
+use slug_types::paths::GardenItemUrl;
 
-use super::helpers::{item_path_for_api, resolve_item};
+use super::helpers::resolve_item;
 
 #[derive(Debug)]
 pub struct ValidatedIngest {
@@ -22,6 +23,10 @@ pub fn validate_ingest_document(
     text: &str,
     scope: &ScopeId,
 ) -> Result<ValidatedIngest, (StatusCode, String, Option<String>)> {
+    let room_wire = match scope {
+        ScopeId::Public => "public",
+        ScopeId::Room(r) => r.as_str(),
+    };
     let public_content = reduced.public();
     let scoped_content = match scope {
         ScopeId::Public => None,
@@ -61,14 +66,14 @@ pub fn validate_ingest_document(
                 let Some(body_text) = body else {
                     return Err((
                         StatusCode::BAD_REQUEST,
-                        format!("item missing body: {}", item_path_for_api(&item)),
+                        format!("item missing body: {}", GardenItemUrl::from_storage_str(&item, room_wire)),
                         Some("items must be declared with bodies, e.g. `~/path/item { ... }`".to_string()),
                     ));
                 };
                 if body_text.trim().is_empty() {
                     return Err((
                         StatusCode::BAD_REQUEST,
-                        format!("item body is empty: {}", item_path_for_api(&item)),
+                        format!("item body is empty: {}", GardenItemUrl::from_storage_str(&item, room_wire)),
                         Some("write at least one sentence inside `{ ... }`".to_string()),
                     ));
                 }
@@ -101,7 +106,7 @@ pub fn validate_ingest_document(
                         let key = CanonicalItemUrl((*it).clone());
                         !defined_in_doc.contains(*it) && !item_exists(&key)
                     })
-                    .map(|it| item_path_for_api(it))
+                    .map(|it| GardenItemUrl::from_storage_str(it, room_wire).into_inner())
                     .collect();
                 if !missing.is_empty() {
                     return Err((
@@ -119,7 +124,7 @@ pub fn validate_ingest_document(
                         let key = CanonicalItemUrl((*it).clone());
                         !defined_in_doc.contains(*it) && !body_exists(&key)
                     })
-                    .map(|it| item_path_for_api(it))
+                    .map(|it| GardenItemUrl::from_storage_str(it, room_wire).into_inner())
                     .collect();
                 if !missing_body.is_empty() {
                     return Err((
diff --git a/server/src/canonical_path.rs b/server/src/canonical_path.rs
index 5c0febe883d8b3978a25896ed0d71df8509a8717..8a1998025121838b0867f8c5ddbd28aea41a23d2 100644
--- a/server/src/canonical_path.rs
+++ b/server/src/canonical_path.rs
@@ -1,92 +1,3 @@
-//! Normalization for thread tags and ontology item URLs (DSL ↔ stored canonical form).
-//! Not event types — see `events` and `path_types`.
+//! Re-exports — implementations live in `slug-types` (`paths` module).
 
-/// Thread / public tag: stored without leading `#`, lowercase.
-pub fn canonicalize_tag(input: &str) -> String {
-    input.trim().trim_start_matches('#').to_lowercase()
-}
-
-/// Ontology item reference → canonical absolute URL on the slug host.
-pub fn canonicalize_item(input: &str) -> String {
-    let s = input.trim();
-    if s.is_empty() {
-        return String::new();
-    }
-
-    if let Some(rest) = s.strip_prefix("https://") {
-        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
-        let host = host.trim().to_lowercase();
-        if tail.is_empty() {
-            return format!("https://{}", host);
-        } else {
-            return format!("https://{}/{}", host, tail);
-        }
-    }
-    if let Some(rest) = s.strip_prefix("http://") {
-        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
-        let host = host.trim().to_lowercase();
-        if tail.is_empty() {
-            return format!("http://{}", host);
-        } else {
-            return format!("http://{}/{}", host, tail);
-        }
-    }
-
-    let is_tilde = s.starts_with("~/");
-    let rest = s.strip_prefix("~/").or_else(|| s.strip_prefix("/")).unwrap_or(s);
-
-    let tail = rest
-        .split('/')
-        .filter_map(|seg| {
-            let t = seg.trim();
-            if t.is_empty() {
-                None
-            } else {
-                Some(t.to_lowercase())
-            }
-        })
-        .collect::<Vec<_>>()
-        .join("/");
-
-    if is_tilde {
-        format!("https://slug.social/~/{}", tail)
-    } else if tail.is_empty() {
-        "https://slug.social".to_string()
-    } else {
-        format!("https://slug.social/{}", tail)
-    }
-}
-
-pub fn item_path_segments(input: &str) -> Vec<String> {
-    let canonical = canonicalize_item(input);
-    if canonical.is_empty() {
-        return vec![];
-    }
-
-    if let Some(rest) = canonical.strip_prefix("https://") {
-        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
-        let mut out = vec![format!("https://{}", host)];
-        out.extend(tail.split('/').filter(|s| !s.is_empty()).map(|s| s.to_string()));
-        return out;
-    }
-    if let Some(rest) = canonical.strip_prefix("http://") {
-        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
-        let mut out = vec![format!("http://{}", host)];
-        out.extend(tail.split('/').filter(|s| !s.is_empty()).map(|s| s.to_string()));
-        return out;
-    }
-
-    canonical
-        .split('/')
-        .filter(|s| !s.is_empty())
-        .map(|s| s.to_string())
-        .collect()
-}
-
-pub fn item_parent_path(input: &str) -> Option<String> {
-    let segs = item_path_segments(input);
-    if segs.len() <= 1 {
-        return None;
-    }
-    Some(segs[..segs.len() - 1].join("/"))
-}
+pub use slug_types::paths::{canonicalize_item, canonicalize_tag, item_parent_path, item_path_segments};
diff --git a/server/src/html/forum.rs b/server/src/html/forum.rs
index ab7970feca47a4431ddc19dab2c8186180b78413..f789e347dc136f8ffd356f4492f0bd76cea04bf0 100644
--- a/server/src/html/forum.rs
+++ b/server/src/html/forum.rs
@@ -697,7 +697,7 @@ async fn thread_view_inner(
     let offset = q.offset.unwrap_or(0);
     let page_ids: Vec<String> = all_ids.into_iter().skip(offset).take(PAGE_SIZE).collect();
 
-    let (display_ingests, subtitle) = {
+    let (display_ingests, _subtitle) = {
         let reduced = state.reduced.read().await;
         let ingests = page_ids
             .iter()
diff --git a/server/src/path_types.rs b/server/src/path_types.rs
index cad58cac7da948f92d6ea2a5587d917ab21d57ea..4c8075bbd488f1b9a5eda9e03ced83f6238c6d5e 100644
--- a/server/src/path_types.rs
+++ b/server/src/path_types.rs
@@ -1,260 +1,3 @@
-//! Path representation types.
-//!
-//! The codebase currently treats item identifiers as strings in a few different
-//! encodings:
-//! - user/DSL input like `~/a/b`
-//! - canonical item URLs like `https://slug.social/~/a/b`
-//! - relative paths within a rooted tree view (e.g. `llms/openai` under a root)
-//!
-//! This module adds lightweight newtypes so code can be explicit about what it
-//! expects without changing core storage formats.
-//!
-//! **Storage vs wire:** [`CanonicalItemUrl`] values are shared across scopes
-//! (`https://slug.social/~/…`); which [`crate::reducer::ContentState`] they live in
-//! is determined by scope, not by embedding the room id in the string. For JSON/RPC
-//! and browser links in a private room, use [`crate::api::helpers::item_path_for_api_in_room`]
-//! so ontology items become `https://slug.social/r/{short}/{slug}/~/…`.
+//! Re-exports — implementations live in `slug-types` (`paths` module).
 
-use std::borrow::Borrow;
-use std::fmt;
-
-use serde::{Deserialize, Serialize};
-
-use crate::canonical_path::canonicalize_item;
-
-/// Canonical item identifier as produced by `canonical_path::canonicalize_item`.
-///
-/// In practice this is usually:
-/// - `https://slug.social/~/...` for ontology items, or
-/// - `https://...` / `http://...` for URL items.
-#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
-pub struct CanonicalItemUrl(pub String);
-
-impl CanonicalItemUrl {
-    pub fn parse(input: &str) -> Option<Self> {
-        let c = canonicalize_item(input);
-        if c.is_empty() {
-            None
-        } else {
-            Some(Self(c))
-        }
-    }
-
-    pub fn as_str(&self) -> &str {
-        &self.0
-    }
-
-    /// Returns the `~/...` tail for ontology items (`https://slug.social/~/...`).
-    pub fn tilde_tail(&self) -> Option<&str> {
-        self.0.strip_prefix("https://slug.social/~/")
-    }
-
-    /// Returns the final non-empty `/`-separated segment of the path.
-    ///
-    /// `https://slug.social/~/a/b/c` → `"c"`
-    /// `https://slug.social/~/a`     → `"a"`
-    pub fn last_segment(&self) -> &str {
-        self.0
-            .rsplit('/')
-            .find(|s| !s.is_empty())
-            .unwrap_or(self.0.as_str())
-    }
-
-    /// The ontology root key as stored in `item_children`: `"https://slug.social/~"`.
-    /// Use this (not `parse("~/")`) when looking up top-level children.
-    pub fn ontology_root() -> Self {
-        Self("https://slug.social/~".to_string())
-    }
-
-    /// Returns the parent of this canonical item URL by stripping the last
-    /// path segment, or `None` if there is no parent (already at root).
-    ///
-    /// `https://slug.social/~/a/b/c` → `Some("https://slug.social/~/a/b")`
-    /// `https://slug.social/~/a`     → `Some("https://slug.social/~")`
-    /// `https://slug.social/~/`      → `None`  (tilde root)
-    pub fn parent(&self) -> Option<Self> {
-        // tilde_tail() is None for non-ontology URLs and "" for the root ~/
-        if self.tilde_tail().map(|t| t.is_empty()).unwrap_or(true) {
-            return None;
-        }
-        // Strip everything from the last '/' onwards.
-        let last_slash = self.0.rfind('/')?;
-        let parent_str = &self.0[..last_slash];
-        if parent_str.is_empty() {
-            None
-        } else {
-            Some(Self(parent_str.to_string()))
-        }
-    }
-
-    /// Segments of an ontology path suitable for breadcrumb rendering.
-    /// Strips the `https://slug.social` prefix and returns the `~/…` parts.
-    ///
-    /// `https://slug.social/~/a/b` → `["~", "a", "b"]`
-    /// `https://slug.social/~/`    → `["~"]`
-    pub fn tilde_segments(&self) -> Vec<&str> {
-        match self.tilde_tail() {
-            Some(tail) if !tail.is_empty() => {
-                std::iter::once("~")
-                    .chain(tail.split('/').filter(|s| !s.is_empty()))
-                    .collect()
-            }
-            Some(_) => vec!["~"],
-            None => vec![],
-        }
-    }
-}
-
-impl fmt::Display for CanonicalItemUrl {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        self.0.fmt(f)
-    }
-}
-
-/// Allow `HashMap<CanonicalItemUrl, _>` to be searched by `&str`.
-impl Borrow<str> for CanonicalItemUrl {
-    fn borrow(&self) -> &str {
-        &self.0
-    }
-}
-
-impl PartialEq<str> for CanonicalItemUrl {
-    fn eq(&self, other: &str) -> bool {
-        self.0 == other
-    }
-}
-
-impl PartialEq<&str> for CanonicalItemUrl {
-    fn eq(&self, other: &&str) -> bool {
-        self.0 == *other
-    }
-}
-
-impl PartialEq<String> for CanonicalItemUrl {
-    fn eq(&self, other: &String) -> bool {
-        &self.0 == other
-    }
-}
-
-/// A `~/...` input path (as used in the DSL and UX).
-///
-/// This is not canonicalized; it is a presentation/input form.
-#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
-pub struct TildePath(pub String);
-
-impl TildePath {
-    pub fn new(input: &str) -> Option<Self> {
-        let s = input.trim();
-        if s.starts_with("~/") && s.len() > 2 {
-            Some(Self(s.to_string()))
-        } else if s == "~/" {
-            Some(Self("~/".to_string()))
-        } else {
-            None
-        }
-    }
-
-    pub fn as_str(&self) -> &str {
-        &self.0
-    }
-
-    pub fn canonicalize(&self) -> Option<CanonicalItemUrl> {
-        CanonicalItemUrl::parse(&self.0)
-    }
-}
-
-impl fmt::Display for TildePath {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        self.0.fmt(f)
-    }
-}
-
-/// A path relative to a chosen root in a tree UI.
-///
-/// This is intended for compact state encodings (blobs). It must be joined to a
-/// root `CanonicalItemUrl` (typically an ontology root) to become a full item.
-#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
-pub struct RelativePath(pub String);
-
-impl RelativePath {
-    pub fn new(input: &str) -> Option<Self> {
-        let s = input.trim().trim_matches('/');
-        if s.is_empty() {
-            Some(Self(String::new()))
-        } else {
-            // Keep this permissive: the DSL parser is the main gatekeeper.
-            Some(Self(s.to_string()))
-        }
-    }
-
-    pub fn as_str(&self) -> &str {
-        &self.0
-    }
-
-    /// Join this relative path under a canonical ontology root
-    /// (`https://slug.social/~/...`) to form a canonical item URL.
-    pub fn join_under_ontology_root(&self, root: &CanonicalItemUrl) -> Option<CanonicalItemUrl> {
-        let base = root.tilde_tail()?;
-        // base is the tail after https://slug.social/~/, e.g. "models" or "models/llms"
-        let joined = if base.is_empty() {
-            if self.0.is_empty() {
-                "~/".to_string()
-            } else {
-                format!("~/{}", self.0)
-            }
-        } else if self.0.is_empty() {
-            format!("~/{}", base)
-        } else {
-            format!("~/{}/{}", base.trim_end_matches('/'), self.0)
-        };
-        CanonicalItemUrl::parse(&joined)
-    }
-}
-
-impl fmt::Display for RelativePath {
-    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
-        self.0.fmt(f)
-    }
-}
-
-
-#[cfg(test)]
-mod tests {
-    use super::*;
-
-    #[test]
-    fn canonical_parent_deep() {
-        let c = CanonicalItemUrl::parse("~/a/b/c").unwrap();
-        assert_eq!(c.parent().unwrap().as_str(), "https://slug.social/~/a/b");
-    }
-
-    #[test]
-    fn canonical_parent_one_level() {
-        let c = CanonicalItemUrl::parse("~/a").unwrap();
-        assert_eq!(c.parent().unwrap().as_str(), "https://slug.social/~");
-    }
-
-    #[test]
-    fn canonical_parent_root_is_none() {
-        let root = CanonicalItemUrl::parse("~/").unwrap();
-        assert!(root.parent().is_none());
-    }
-
-    #[test]
-    fn tilde_segments_deep() {
-        let c = CanonicalItemUrl::parse("~/a/b").unwrap();
-        assert_eq!(c.tilde_segments(), vec!["~", "a", "b"]);
-    }
-
-    #[test]
-    fn tilde_segments_root() {
-        let c = CanonicalItemUrl::parse("~/").unwrap();
-        assert_eq!(c.tilde_segments(), vec!["~"]);
-    }
-
-    #[test]
-    fn tilde_segments_non_ontology_is_empty() {
-        let c = CanonicalItemUrl::parse("https://example.com/foo").unwrap();
-        assert_eq!(c.tilde_segments(), Vec::<&str>::new());
-    }
-}
+pub use slug_types::paths::{CanonicalItemUrl, RelativePath, TildePath};
diff --git a/types/src/lib.rs b/types/src/lib.rs
index c1cde3b783d03b02783385b5f07659fb112aae3e..5fc867bf2af84c2ca63fa1cdd03413110ad784a3 100644
--- a/types/src/lib.rs
+++ b/types/src/lib.rs
@@ -1,7 +1,13 @@
 use serde::{Deserialize, Serialize};
 
+pub mod paths;
 pub mod timeago;
 
+pub use paths::{
+    canonicalize_item, canonicalize_tag, item_parent_path, item_path_segments, CanonicalItemUrl,
+    ForumThreadUrl, GardenItemUrl, RelativePath, TildeOntologyPath, TildePath,
+};
+
 #[derive(Debug, Serialize, Deserialize)]
 pub struct ApiError {
     pub ok: bool,
@@ -12,7 +18,7 @@ pub struct ApiError {
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct RankRow {
-    pub item: String,
+    pub item: GardenItemUrl,
     pub score: f64,
     /// Normalized score as a percentage of the top item (0–100). Present when ?percent=true.
     #[serde(skip_serializing_if = "Option::is_none")]
@@ -43,7 +49,7 @@ pub struct RankComponent {
 #[derive(Debug, Serialize, Deserialize)]
 pub struct RankResponse {
     pub components: Vec<RankComponent>,
-    pub unranked_items: Vec<String>,
+    pub unranked_items: Vec<GardenItemUrl>,
 }
 
 /// Graph connectivity stats for a scope, returned with pair suggestions.
@@ -63,8 +69,8 @@ pub struct ConnectivityStats {
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct PairResponse {
-    pub left: String,
-    pub right: String,
+    pub left: GardenItemUrl,
+    pub right: GardenItemUrl,
     pub left_body: Option<String>,
     pub right_body: Option<String>,
     /// Thread tags that discuss either item (connective tissue to forum).
@@ -79,7 +85,7 @@ pub struct PairResponse {
 pub struct NextMoves {
     pub pair: String,
     pub rank: String,
-    pub web: String,
+    pub web: ForumThreadUrl,
 }
 
 #[derive(Debug, Serialize, Deserialize)]
@@ -90,14 +96,14 @@ pub struct PathsResponse {
 /// Leaf items only (no children). For search / "full path list" — does not scale, works for now.
 #[derive(Debug, Serialize, Deserialize)]
 pub struct LeavesResponse {
-    pub paths: Vec<String>,
+    pub paths: Vec<GardenItemUrl>,
 }
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct PathSummary {
-    pub path: String,
+    pub path: TildeOntologyPath,
     pub children: usize,
-    pub web: String,
+    pub web: GardenItemUrl,
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
@@ -109,7 +115,7 @@ pub struct ThreadsResponse {
 pub struct ThreadSummary {
     pub thread: String,
     pub last_activity_ts: i64,
-    pub web: String,
+    pub web: ForumThreadUrl,
 }
 
 #[derive(Debug, Serialize, Deserialize)]
@@ -178,7 +184,7 @@ pub struct IngestRow {
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct ItemResponse {
-    pub item: String,
+    pub item: GardenItemUrl,
     pub body: Option<String>,
     /// True when the body was truncated due to size. Fetch with `?full=true` for the complete body.
     #[serde(default, skip_serializing_if = "std::ops::Not::not")]
@@ -201,15 +207,15 @@ pub struct RecentVotesResponse {
 /// Vote history for one item (matchup: wins/losses + thread per vote).
 #[derive(Debug, Serialize, Deserialize)]
 pub struct MatchupResponse {
-    pub item: String,
+    pub item: GardenItemUrl,
     pub votes: Vec<VoteRow>,
 }
 
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct VoteRow {
     pub ts: i64,
-    pub a: String,
-    pub b: String,
+    pub a: GardenItemUrl,
+    pub b: GardenItemUrl,
     pub ratio: String,
     /// Principal username when present (stored form, no `@`).
     pub actor: Option<String>,
@@ -510,7 +516,7 @@ pub struct RankPosition {
 /// How one item's rank changed after a vote.
 #[derive(Debug, Clone, Serialize, Deserialize)]
 pub struct RankChange {
-    pub item: String,
+    pub item: GardenItemUrl,
     /// Position before the vote. None = was unranked (no voted connections in this scope).
     pub before: Option<RankPosition>,
     /// Position after the vote. None = became unranked (e.g. component split, unlikely).
@@ -544,7 +550,7 @@ pub struct CheckScopeRanking {
     /// Parent scope path (e.g. "/models" or "/" for root).
     pub parent: String,
     pub components: Vec<RankComponent>,
-    pub unranked_items: Vec<String>,
+    pub unranked_items: Vec<GardenItemUrl>,
 }
 
 #[derive(Debug, Serialize, Deserialize)]
@@ -567,7 +573,7 @@ pub struct SearchResponse {
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct SearchItemHit {
-    pub path: String,
+    pub path: GardenItemUrl,
     #[serde(skip_serializing_if = "Option::is_none")]
     pub body: Option<String>,
 }
@@ -614,7 +620,7 @@ pub struct RankHistoryRow {
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct RankHistoryResponse {
-    pub item: String,
+    pub item: GardenItemUrl,
     pub history: Vec<RankHistoryRow>,
 }
 
diff --git a/types/src/paths.rs b/types/src/paths.rs
new file mode 100644
index 0000000000000000000000000000000000000000..2950a7502255927583fbacdfd2adb700f0b0c221
--- /dev/null
+++ b/types/src/paths.rs
@@ -0,0 +1,498 @@
+//! Canonical paths, storage ids, and JSON href newtypes. All normalization and
+//! room-aware URL rules for items live here.
+
+use std::borrow::Borrow;
+use std::fmt;
+
+use serde::{Deserialize, Serialize};
+
+// ---------------------------------------------------------------------------
+// Normalization (moved from server `canonical_path`)
+// ---------------------------------------------------------------------------
+
+/// Thread / public tag: stored without leading `#`, lowercase.
+pub fn canonicalize_tag(input: &str) -> String {
+    input.trim().trim_start_matches('#').to_lowercase()
+}
+
+/// Ontology item reference → canonical absolute URL on the slug host.
+pub fn canonicalize_item(input: &str) -> String {
+    let s = input.trim();
+    if s.is_empty() {
+        return String::new();
+    }
+
+    if let Some(rest) = s.strip_prefix("https://") {
+        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
+        let host = host.trim().to_lowercase();
+        if tail.is_empty() {
+            return format!("https://{}", host);
+        } else {
+            return format!("https://{}/{}", host, tail);
+        }
+    }
+    if let Some(rest) = s.strip_prefix("http://") {
+        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
+        let host = host.trim().to_lowercase();
+        if tail.is_empty() {
+            return format!("http://{}", host);
+        } else {
+            return format!("http://{}/{}", host, tail);
+        }
+    }
+
+    let is_tilde = s.starts_with("~/");
+    let rest = s.strip_prefix("~/").or_else(|| s.strip_prefix("/")).unwrap_or(s);
+
+    let tail = rest
+        .split('/')
+        .filter_map(|seg| {
+            let t = seg.trim();
+            if t.is_empty() {
+                None
+            } else {
+                Some(t.to_lowercase())
+            }
+        })
+        .collect::<Vec<_>>()
+        .join("/");
+
+    if is_tilde {
+        format!("https://slug.social/~/{}", tail)
+    } else if tail.is_empty() {
+        "https://slug.social".to_string()
+    } else {
+        format!("https://slug.social/{}", tail)
+    }
+}
+
+pub fn item_path_segments(input: &str) -> Vec<String> {
+    let canonical = canonicalize_item(input);
+    if canonical.is_empty() {
+        return vec![];
+    }
+
+    if let Some(rest) = canonical.strip_prefix("https://") {
+        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
+        let mut out = vec![format!("https://{}", host)];
+        out.extend(tail.split('/').filter(|s| !s.is_empty()).map(|s| s.to_string()));
+        return out;
+    }
+    if let Some(rest) = canonical.strip_prefix("http://") {
+        let (host, tail) = rest.split_once('/').map_or((rest, ""), |(h, t)| (h, t));
+        let mut out = vec![format!("http://{}", host)];
+        out.extend(tail.split('/').filter(|s| !s.is_empty()).map(|s| s.to_string()));
+        return out;
+    }
+
+    canonical
+        .split('/')
+        .filter(|s| !s.is_empty())
+        .map(|s| s.to_string())
+        .collect()
+}
+
+pub fn item_parent_path(input: &str) -> Option<String> {
+    let segs = item_path_segments(input);
+    if segs.len() <= 1 {
+        return None;
+    }
+    Some(segs[..segs.len() - 1].join("/"))
+}
+
+// ---------------------------------------------------------------------------
+// Storage + input path newtypes
+// ---------------------------------------------------------------------------
+
+/// Canonical item identifier as produced by [`canonicalize_item`].
+///
+/// Shared across all scopes; room is not embedded. Usually
+/// `https://slug.social/~/…` or an external `http(s)://…` URL item.
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
+pub struct CanonicalItemUrl(pub String);
+
+impl CanonicalItemUrl {
+    pub fn parse(input: &str) -> Option<Self> {
+        let c = canonicalize_item(input);
+        if c.is_empty() {
+            None
+        } else {
+            Some(Self(c))
+        }
+    }
+
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+
+    pub fn tilde_tail(&self) -> Option<&str> {
+        self.0.strip_prefix("https://slug.social/~/")
+    }
+
+    pub fn last_segment(&self) -> &str {
+        self.0
+            .rsplit('/')
+            .find(|s| !s.is_empty())
+            .unwrap_or(self.0.as_str())
+    }
+
+    pub fn ontology_root() -> Self {
+        Self("https://slug.social/~".to_string())
+    }
+
+    pub fn parent(&self) -> Option<Self> {
+        if self.tilde_tail().map(|t| t.is_empty()).unwrap_or(true) {
+            return None;
+        }
+        let last_slash = self.0.rfind('/')?;
+        let parent_str = &self.0[..last_slash];
+        if parent_str.is_empty() {
+            None
+        } else {
+            Some(Self(parent_str.to_string()))
+        }
+    }
+
+    pub fn tilde_segments(&self) -> Vec<&str> {
+        match self.tilde_tail() {
+            Some(tail) if !tail.is_empty() => {
+                std::iter::once("~")
+                    .chain(tail.split('/').filter(|s| !s.is_empty()))
+                    .collect()
+            }
+            Some(_) => vec!["~"],
+            None => vec![],
+        }
+    }
+
+    /// `~/…` list label for ontology items (paths index, CLI).
+    pub fn tilde_list_label(&self) -> TildeOntologyPath {
+        TildeOntologyPath::from_stored(self)
+    }
+
+    /// Absolute href for JSON/RPC and browsers for this stored id in `room`.
+    pub fn json_href(&self, room_wire: &str) -> GardenItemUrl {
+        GardenItemUrl::from_stored(self, room_wire)
+    }
+}
+
+impl fmt::Display for CanonicalItemUrl {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+impl Borrow<str> for CanonicalItemUrl {
+    fn borrow(&self) -> &str {
+        &self.0
+    }
+}
+
+impl PartialEq<str> for CanonicalItemUrl {
+    fn eq(&self, other: &str) -> bool {
+        self.0 == other
+    }
+}
+
+impl PartialEq<&str> for CanonicalItemUrl {
+    fn eq(&self, other: &&str) -> bool {
+        self.0 == *other
+    }
+}
+
+impl PartialEq<String> for CanonicalItemUrl {
+    fn eq(&self, other: &String) -> bool {
+        &self.0 == other
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
+pub struct TildePath(pub String);
+
+impl TildePath {
+    pub fn new(input: &str) -> Option<Self> {
+        let s = input.trim();
+        if s.starts_with("~/") && s.len() > 2 {
+            Some(Self(s.to_string()))
+        } else if s == "~/" {
+            Some(Self("~/".to_string()))
+        } else {
+            None
+        }
+    }
+
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+
+    pub fn canonicalize(&self) -> Option<CanonicalItemUrl> {
+        CanonicalItemUrl::parse(&self.0)
+    }
+}
+
+impl fmt::Display for TildePath {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
+pub struct RelativePath(pub String);
+
+impl RelativePath {
+    pub fn new(input: &str) -> Option<Self> {
+        let s = input.trim().trim_matches('/');
+        if s.is_empty() {
+            Some(Self(String::new()))
+        } else {
+            Some(Self(s.to_string()))
+        }
+    }
+
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+
+    pub fn join_under_ontology_root(&self, root: &CanonicalItemUrl) -> Option<CanonicalItemUrl> {
+        let base = root.tilde_tail()?;
+        let joined = if base.is_empty() {
+            if self.0.is_empty() {
+                "~/".to_string()
+            } else {
+                format!("~/{}", self.0)
+            }
+        } else if self.0.is_empty() {
+            format!("~/{}", base)
+        } else {
+            format!("~/{}/{}", base.trim_end_matches('/'), self.0)
+        };
+        CanonicalItemUrl::parse(&joined)
+    }
+}
+
+impl fmt::Display for RelativePath {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+// ---------------------------------------------------------------------------
+// Wire / JSON: correct-by-construction hrefs
+// ---------------------------------------------------------------------------
+
+fn api_path_or_url(item: &str) -> String {
+    if item.starts_with("http://") || item.starts_with("https://") {
+        item.to_string()
+    } else {
+        format!("/{}", item)
+    }
+}
+
+/// Ontology item as serialized in JSON (absolute URL or `/`-prefixed path).
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct GardenItemUrl(pub String);
+
+impl GardenItemUrl {
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+
+    pub fn into_inner(self) -> String {
+        self.0
+    }
+
+    /// Stored canonical id + RPC `room` field (`"public"` or `"short/slug"`).
+    pub fn from_stored(stored: &CanonicalItemUrl, room_wire: &str) -> Self {
+        Self(garden_href_string(stored.as_str(), room_wire))
+    }
+
+    /// Like [`Self::from_stored`] but accepts a string that may already be canonical.
+    pub fn from_storage_str(stored: &str, room_wire: &str) -> Self {
+        Self(garden_href_string(stored, room_wire))
+    }
+}
+
+impl fmt::Display for GardenItemUrl {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+fn garden_href_string(item: &str, room_wire: &str) -> String {
+    let room = room_wire.trim();
+    if room.is_empty() || room == "public" {
+        return api_path_or_url(item);
+    }
+    let Some((short, slug)) = room.split_once('/') else {
+        return api_path_or_url(item);
+    };
+    if short.is_empty() || slug.is_empty() {
+        return api_path_or_url(item);
+    }
+    let Some(c) = CanonicalItemUrl::parse(item) else {
+        return api_path_or_url(item);
+    };
+    let root = CanonicalItemUrl::ontology_root();
+    let item_norm = c.as_str().trim_end_matches('/');
+    let root_norm = root.as_str().trim_end_matches('/');
+    if let Some(tail) = c.tilde_tail() {
+        return if tail.is_empty() {
+            format!("https://slug.social/r/{short}/{slug}/~")
+        } else {
+            format!("https://slug.social/r/{short}/{slug}/~/{}", tail)
+        };
+    }
+    if item_norm == root_norm {
+        return format!("https://slug.social/r/{short}/{slug}/~");
+    }
+    api_path_or_url(item)
+}
+
+/// Forum thread URL for JSON (`/t/…` or `/r/…/t/…` on slug.social).
+#[derive(Debug, Clone, PartialEq, Eq, PartialOrd, Ord, Hash, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct ForumThreadUrl(pub String);
+
+impl ForumThreadUrl {
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+
+    pub fn into_inner(self) -> String {
+        self.0
+    }
+
+    pub fn from_room_tag(room_wire: &str, thread_tag: &str) -> Self {
+        let room = room_wire.trim();
+        let tag = thread_tag.trim().trim_start_matches('#');
+        Self(if room.is_empty() || room == "public" {
+            format!("https://slug.social/t/{tag}")
+        } else if let Some((short, slug)) = room.split_once('/') {
+            if short.is_empty() || slug.is_empty() {
+                format!("https://slug.social/t/{tag}")
+            } else {
+                format!("https://slug.social/r/{short}/{slug}/t/{tag}")
+            }
+        } else {
+            format!("https://slug.social/t/{tag}")
+        })
+    }
+}
+
+impl fmt::Display for ForumThreadUrl {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+/// `~/a/b` style path for list UIs (paths index `path` field).
+#[derive(Debug, Clone, PartialEq, Eq, Hash, Serialize, Deserialize)]
+#[serde(transparent)]
+pub struct TildeOntologyPath(pub String);
+
+impl TildeOntologyPath {
+    pub fn from_stored(c: &CanonicalItemUrl) -> Self {
+        let s = match c.tilde_tail() {
+            Some(tail) if !tail.is_empty() => format!("~/{}", tail),
+            Some(_) => "~/".to_string(),
+            None => c.to_string(),
+        };
+        Self(s)
+    }
+
+    pub fn as_str(&self) -> &str {
+        &self.0
+    }
+}
+
+impl fmt::Display for TildeOntologyPath {
+    fn fmt(&self, f: &mut fmt::Formatter<'_>) -> fmt::Result {
+        self.0.fmt(f)
+    }
+}
+
+#[cfg(test)]
+mod tests {
+    use super::*;
+
+    #[test]
+    fn canonical_parent_deep() {
+        let c = CanonicalItemUrl::parse("~/a/b/c").unwrap();
+        assert_eq!(c.parent().unwrap().as_str(), "https://slug.social/~/a/b");
+    }
+
+    #[test]
+    fn canonical_parent_one_level() {
+        let c = CanonicalItemUrl::parse("~/a").unwrap();
+        assert_eq!(c.parent().unwrap().as_str(), "https://slug.social/~");
+    }
+
+    #[test]
+    fn canonical_parent_root_is_none() {
+        let root = CanonicalItemUrl::parse("~/").unwrap();
+        assert!(root.parent().is_none());
+    }
+
+    #[test]
+    fn tilde_segments_deep() {
+        let c = CanonicalItemUrl::parse("~/a/b").unwrap();
+        assert_eq!(c.tilde_segments(), vec!["~", "a", "b"]);
+    }
+
+    #[test]
+    fn tilde_segments_root() {
+        let c = CanonicalItemUrl::parse("~/").unwrap();
+        assert_eq!(c.tilde_segments(), vec!["~"]);
+    }
+
+    #[test]
+    fn tilde_segments_non_ontology_is_empty() {
+        let c = CanonicalItemUrl::parse("https://example.com/foo").unwrap();
+        assert_eq!(c.tilde_segments(), Vec::<&str>::new());
+    }
+
+    #[test]
+    fn garden_public_passthrough_https() {
+        let u = "https://slug.social/~/a/b";
+        assert_eq!(GardenItemUrl::from_storage_str(u, "public").as_str(), u);
+    }
+
+    #[test]
+    fn garden_private_room_prefixes_ontology() {
+        assert_eq!(
+            GardenItemUrl::from_storage_str("https://slug.social/~/topic/x", "9ab12cd/my-room").as_str(),
+            "https://slug.social/r/9ab12cd/my-room/~/topic/x"
+        );
+    }
+
+    #[test]
+    fn garden_private_room_ontology_root() {
+        assert_eq!(
+            GardenItemUrl::from_storage_str("https://slug.social/~", "9ab12cd/my-room").as_str(),
+            "https://slug.social/r/9ab12cd/my-room/~"
+        );
+        assert_eq!(
+            GardenItemUrl::from_storage_str("https://slug.social/~/", "9ab12cd/my-room").as_str(),
+            "https://slug.social/r/9ab12cd/my-room/~"
+        );
+    }
+
+    #[test]
+    fn garden_external_url_untouched_in_private_room() {
+        let u = "https://example.com/z";
+        assert_eq!(GardenItemUrl::from_storage_str(u, "9ab12cd/my-room").as_str(), u);
+    }
+
+    #[test]
+    fn forum_web_public_vs_room() {
+        assert_eq!(
+            ForumThreadUrl::from_room_tag("public", "debate").as_str(),
+            "https://slug.social/t/debate"
+        );
+        assert_eq!(
+            ForumThreadUrl::from_room_tag("9ab12cd/my-room", "#debate").as_str(),
+            "https://slug.social/r/9ab12cd/my-room/t/debate"
+        );
+    }
+}


Side B — contributor: tommy-mor
Side B — commit message:
[6d04afc2] refactor

Side B — unified diff (full patch):
diff --git a/Cargo.lock b/Cargo.lock
index 2cea973082716e761ef6f5dd5886acc08ff9aac0..8c43fb75c472b102e6e1d3b837dce3355be898f2 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -17,6 +17,28 @@ version = "1.0.102"
 source = "registry+https://github.com/rust-lang/crates.io-index"
 checksum = "7f202df86484c868dbad7eaa557ef785d5c66295e41b460ef922eca0723b842c"
 
+[[package]]
+name = "async-stream"
+version = "0.3.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "0b5a71a6f37880a80d1d7f19efd781e4b5de42c88f0722cc13bcb6cc2cfe8476"
+dependencies = [
+ "async-stream-impl",
+ "futures-core",
+ "pin-project-lite",
+]
+
+[[package]]
+name = "async-stream-impl"
+version = "0.3.6"
+source = "registry+https://github.com/rust-lang/crates.io-index"
+checksum = "c7c24de15d275a1ecfd47a380fb4d5ec9bfe0933f309ed5e705b775596a3574d"
+dependencies = [
+ "proc-macro2",
+ "quote",
+ "syn",
+]
+
 [[package]]
 name = "async-trait"
 version = "0.1.89"
@@ -1242,9 +1264,11 @@ dependencies = [
 name = "sorter2-server"
 version = "0.0.1"
 dependencies = [
+ "async-stream",
  "axum",
  "axum-extra",
  "dotenvy",
+ "futures-util",
  "maud",
  "reqwest",
  "serde",
diff --git a/server/Cargo.toml b/server/Cargo.toml
index bd600138b613bd0f546bdec217a5334cdcb20aa5..c940acb687fb141d21760a3d6656172013cf6f41 100644
--- a/server/Cargo.toml
+++ b/server/Cargo.toml
@@ -18,6 +18,8 @@ tracing = "0.1"
 tracing-subscriber = { version = "0.3", features = ["env-filter"] }
 reqwest = { version = "0.12", features = ["json"] }
 dotenvy = "0.15"
+async-stream = "0.3"
+futures-util = { version = "0.3", default-features = false, features = ["std"] }
 
 [dev-dependencies]
 reqwest = { version = "0.12", features = ["json"] }
diff --git a/server/src/api/ui_html.rs b/server/src/api/ui_html.rs
index b33a84e8bb5e817b26592868d88090e6d664d950..7af6527d03c483f33f3469ce6766c01a554c5fe3 100644
--- a/server/src/api/ui_html.rs
+++ b/server/src/api/ui_html.rs
@@ -6,7 +6,8 @@ use axum::{
 use std::collections::HashMap;
 
 use crate::{
-    html::{entity_section, input_panel, js_string_literal, ranking_panel, JsBuilder},
+    fetch,
+    html::{input_panel, js_string_literal, ranking_panel, JsBuilder},
     parser::parse_reddit_url,
     path_types::ItemId,
     reddit::ensure_partial_tree,
@@ -89,18 +90,8 @@ pub async fn post_ui_html(
         },
         HtmlUiAction::FetchEntity { item } => {
             let id = parse_item_param(&item);
-            if id.is_root() {
-                return ui_js_warn("nothing to fetch for the root").into_response();
-            }
-            state.queue_entity_fetch(id.clone());
-            let tree = state.tree.read().await;
-            let empty = crate::reducer::NodeState::default();
-            let node = tree.get(&id).unwrap_or(&empty);
-            let panel = entity_section(&id, node, true);
-            JsBuilder::new()
-                .morph_selector("#entity-section", panel)
-                .into_response()
-        },
+            fetch::fetch_entity_stream(state, id).into_response()
+        }
     }
 }
 
diff --git a/server/src/fetch/html.rs b/server/src/fetch/html.rs
new file mode 100644
index 0000000000000000000000000000000000000000..63634508496e224c38b9ec0308b7a6086462f925
--- /dev/null
+++ b/server/src/fetch/html.rs
@@ -0,0 +1,67 @@
+//! Markup for entity import / “Fetch from Reddit” (`POST /ui`, SSE response).
+
+use maud::{html, Markup};
+
+use crate::{
+    form_template::template_json_compact,
+    path_types::ItemId,
+    reddit::is_fetchable,
+    reducer::NodeState,
+    ui_action::UI_RPC_FIELD,
+};
+
+fn entity_panel(node: &NodeState) -> Markup {
+    html! {
+        @if let Some(data) = &node.data {
+            div id="entity-panel" class="entity-card" {
+                h2 { (data.title) }
+                @if let Some(author) = &data.author {
+                    p class="muted small" { "by " (author) }
+                }
+                @if let Some(body) = &data.body_html {
+                    div class="entity-body" { (maud::PreEscaped(body)) }
+                }
+            }
+        }
+    }
+}
+
+/// Reddit/API import — `POST /ui` with `fetch_entity` returns an SSE stream.
+pub fn fetch_entity_panel(item: &ItemId, has_data: bool, fetching: bool) -> Markup {
+    if !is_fetchable(item) {
+        return html! {};
+    }
+    let label = if fetching {
+        "Fetching…"
+    } else if has_data {
+        "Fetch more"
+    } else {
+        "Fetch from Reddit"
+    };
+    let rpc = template_json_compact(&serde_json::json!({
+        "action": "fetch_entity",
+        "item": item.as_str(),
+    }))
+    .expect("fetch_entity rpc template");
+    html! {
+        form method="post" action="/ui" id="fetch-entity-form" class="fetch-entity-form" {
+            input type="hidden" name=(UI_RPC_FIELD) value=(rpc);
+            @if fetching {
+                button type="submit" class="btn-secondary" disabled { (label) }
+            } @else {
+                button type="submit" class="btn-secondary" { (label) }
+            }
+        }
+    }
+}
+
+/// Entity card + fetch control (target `#entity-section` for Idiomorph / SSE).
+pub fn entity_section(item: &ItemId, node: &NodeState, fetching: bool) -> Markup {
+    let has_data = node.data.is_some();
+    html! {
+        section id="entity-section" class="demo-panel" {
+            (entity_panel(node))
+            (fetch_entity_panel(item, has_data, fetching))
+        }
+    }
+}
diff --git a/server/src/fetch/mod.rs b/server/src/fetch/mod.rs
new file mode 100644
index 0000000000000000000000000000000000000000..2290f9d3a0f1cbf1806c6339f82a4515c11cc3d3
--- /dev/null
+++ b/server/src/fetch/mod.rs
@@ -0,0 +1,115 @@
+//! Entity import over `POST /ui` as SSE (Reddit worker in [`crate::reddit`]).
+
+pub mod html;
+
+use std::convert::Infallible;
+use std::time::Duration;
+
+use async_stream::stream;
+use axum::response::sse::{Event, KeepAlive, Sse};
+use futures_util::Stream;
+use serde::Serialize;
+use tokio::sync::oneshot;
+
+use crate::{
+    path_types::ItemId,
+    reddit::FetchJobResult,
+    reducer::NodeState,
+    state::AppState,
+};
+
+pub fn now_ms() -> i64 {
+    let t = std::time::SystemTime::now()
+        .duration_since(std::time::UNIX_EPOCH)
+        .unwrap_or_default();
+    t.as_millis() as i64
+}
+
+#[derive(Serialize)]
+struct SseMorphPayload {
+    selector: &'static str,
+    html: String,
+}
+
+fn morph_complete_event(html: maud::Markup) -> Event {
+    let payload = SseMorphPayload {
+        selector: "#entity-section",
+        html: html.into_string(),
+    };
+    let data = serde_json::to_string(&payload).unwrap_or_else(|_| "{}".into());
+    Event::default().event("complete").data(data)
+}
+
+/// Stream `fetching` → `complete` / `error` for [`crate::ui_action::HtmlUiAction::FetchEntity`].
+pub fn fetch_entity_stream(
+    state: AppState,
+    id: ItemId,
+) -> Sse<impl Stream<Item = Result<Event, Infallible>>> {
+    tracing::debug!(item = %id, "fetch entity stream opened");
+
+    let stream = stream! {
+        if id.is_root() {
+            yield Ok(Event::default().event("error").data("{\"message\":\"nothing to fetch for the root\"}"));
+            return;
+        }
+
+        if !crate::reddit::is_fetchable(&id) {
+            tracing::debug!(item = %id, "fetch stream: not fetchable");
+            yield Ok(Event::default().event("error").data("{\"message\":\"this page cannot be fetched from Reddit\"}"));
+            return;
+        }
+
+        let fetching_html = {
+            let tree = state.tree.read().await;
+            let empty = NodeState::default();
+            let node = tree.get(&id).unwrap_or(&empty);
+            html::entity_section(&id, node, true).into_string()
+        };
+        let fetching_payload = serde_json::json!({
+            "selector": "#entity-section",
+            "html": fetching_html,
+        });
+        yield Ok(Event::default().event("fetching").data(fetching_payload.to_string()));
+
+        let (tx, rx) = oneshot::channel();
+        state.reddit.request_fetch(id.clone(), true, Some(tx));
+        tracing::debug!(item = %id, "fetch stream: queued reddit job");
+
+        let result = match rx.await {
+            Ok(r) => r,
+            Err(_) => {
+                tracing::warn!(item = %id, "fetch stream: worker dropped oneshot");
+                FetchJobResult::Failed("reddit worker stopped".into())
+            }
+        };
+
+        tracing::debug!(item = %id, ?result, "fetch stream: job finished");
+
+        match result {
+            FetchJobResult::Imported | FetchJobResult::NotFound => {
+                let tree = state.tree.read().await;
+                let empty = NodeState::default();
+                let node = tree.get(&id).unwrap_or(&empty);
+                yield Ok(morph_complete_event(html::entity_section(&id, node, false)));
+            }
+            FetchJobResult::SkippedCached | FetchJobResult::SkippedDuplicate => {
+                let tree = state.tree.read().await;
+                let empty = NodeState::default();
+                let node = tree.get(&id).unwrap_or(&empty);
+                yield Ok(morph_complete_event(html::entity_section(&id, node, false)));
+            }
+            FetchJobResult::RateLimited { reset_secs } => {
+                yield Ok(Event::default().event("error").data(
+                    serde_json::json!({"message": format!("Reddit rate limit — retry in {reset_secs}s")}).to_string(),
+                ));
+            }
+            FetchJobResult::Failed(msg) => {
+                yield Ok(Event::default().event("error").data(
+                    serde_json::json!({"message": msg}).to_string(),
+                ));
+            }
+        }
+    };
+
+    Sse::new(stream).keep_alive(KeepAlive::new().interval(Duration::from_secs(15)))
+}
diff --git a/server/src/html/mod.rs b/server/src/html/mod.rs
index db5b4c7f06b0be64603981166835cde268234f67..9314a7556306ddab969b896dbf4126b542a46722 100644
--- a/server/src/html/mod.rs
+++ b/server/src/html/mod.rs
@@ -7,10 +7,10 @@ use axum::{
 use maud::{html, Markup, DOCTYPE};
 
 use crate::{
+    fetch::html::entity_section,
     form_template::template_json_compact,
     path_types::ItemId,
     ranking::{top_bottom, RankedItem},
-    reddit::is_fetchable,
     reducer::{GroupState, NodeState},
     state::AppState,
     ui_action::UI_RPC_FIELD,
@@ -149,62 +149,6 @@ pub fn breadcrumb_path(item: &ItemId) -> Markup {
     }
 }
 
-fn entity_panel(node: &NodeState) -> Markup {
-    html! {
-        @if let Some(data) = &node.data {
-            div id="entity-panel" class="entity-card" {
-                h2 { (data.title) }
-                @if let Some(author) = &data.author {
-                    p class="muted small" { "by " (author) }
-                }
-                @if let Some(body) = &data.body_html {
-                    div class="entity-body" { (maud::PreEscaped(body)) }
-                }
-            }
-        }
-    }
-}
-
-/// Reddit/API import control — only shown on fetchable pages; never auto-fires.
-pub fn fetch_entity_panel(item: &ItemId, has_data: bool, fetching: bool) -> Markup {
-    if !is_fetchable(item) {
-        return html! {};
-    }
-    let label = if fetching {
-        "Fetching…"
-    } else if has_data {
-        "Fetch more"
-    } else {
-        "Fetch from Reddit"
-    };
-    let rpc = template_json_compact(&serde_json::json!({
-        "action": "fetch_entity",
-        "item": item.as_str(),
-    }))
-    .expect("fetch_entity rpc template");
-    html! {
-        form method="post" action="/ui" id="fetch-entity-form" class="fetch-entity-form" {
-            input type="hidden" name=(UI_RPC_FIELD) value=(rpc);
-            @if fetching {
-                button type="submit" class="btn-secondary" disabled { (label) }
-            } @else {
-                button type="submit" class="btn-secondary" { (label) }
-            }
-        }
-    }
-}
-
-/// Entity card + explicit fetch control (morphed as `#entity-section`).
-pub fn entity_section(item: &ItemId, node: &NodeState, fetching: bool) -> Markup {
-    let has_data = node.data.is_some();
-    html! {
-        section id="entity-section" class="demo-panel" {
-            (entity_panel(node))
-            (fetch_entity_panel(item, has_data, fetching))
-        }
-    }
-}
-
 fn rank_list(label: &str, items: &[RankedItem], start_rank: usize) -> Markup {
     html! {
         @if !items.is_empty() {
diff --git a/server/src/lib.rs b/server/src/lib.rs
index 79b173f391a96ae5d0d96fd656e1e8d2dd070d09..0677363e0ed21d244bfa065f7ec728549ddd9e50 100644
--- a/server/src/lib.rs
+++ b/server/src/lib.rs
@@ -1,6 +1,7 @@
 pub mod api;
 pub mod event_log;
 pub mod events;
+pub mod fetch;
 pub mod form_template;
 pub mod html;
 pub mod parser;
diff --git a/server/src/reddit.rs b/server/src/reddit.rs
index ff0f01e57b18af878eb5be3efc47204a7673589d..0e6ce32720951a58456852b67fb76a85dd9f4aad 100644
--- a/server/src/reddit.rs
+++ b/server/src/reddit.rs
@@ -7,12 +7,12 @@ use std::time::{Duration, Instant};
 use reqwest::{header, Client, StatusCode};
 use serde::Deserialize;
 use serde_json::Value;
-use tokio::sync::{mpsc, RwLock};
+use tokio::sync::{mpsc, oneshot, RwLock};
 
 use crate::{
     event_log::EventLog,
     events::Event,
-    html::now_ms,
+    fetch::now_ms,
     path_types::ItemId,
     reducer::GlobalTree,
 };
@@ -22,10 +22,21 @@ pub fn ensure_partial_tree(tree: &mut GlobalTree, id: &ItemId) {
     tree.ensure_path(id);
 }
 
+#[derive(Debug, Clone, PartialEq, Eq)]
+pub enum FetchJobResult {
+    Imported,
+    NotFound,
+    SkippedDuplicate,
+    SkippedCached,
+    RateLimited { reset_secs: u64 },
+    Failed(String),
+}
+
 pub struct RedditCommand {
     pub id: ItemId,
     /// User-initiated fetch bypasses the in-memory "recently fetched" cache.
     pub force: bool,
+    pub done: Option<oneshot::Sender<FetchJobResult>>,
 }
 
 #[derive(Clone)]
@@ -72,14 +83,29 @@ impl RedditBroker {
             .build()
             .expect("reqwest client");
 
+        tracing::debug!(
+            api_base = %config.api_base,
+            oauth_base = %config.oauth_base,
+            oauth = config.creds.is_some(),
+            "reddit worker started"
+        );
+
         tokio::spawn(reddit_worker(rx, tree, event_log, client, config));
 
         Self { tx }
     }
 
     /// Queue a fetch; drops when the channel is full (backpressure).
-    pub fn request_fetch(&self, id: ItemId, force: bool) {
-        let _ = self.tx.try_send(RedditCommand { id, force });
+    pub fn request_fetch(
+        &self,
+        id: ItemId,
+        force: bool,
+        done: Option<oneshot::Sender<FetchJobResult>>,
+    ) {
+        match self.tx.try_send(RedditCommand { id: id.clone(), force, done }) {
+            Ok(()) => tracing::debug!(item = %id, force, "reddit fetch queued"),
+            Err(_) => tracing::warn!(item = %id, "reddit fetch queue full, dropped"),
+        }
     }
 }
 
@@ -95,8 +121,6 @@ impl RedditApiConfig {
 }
 
 impl RedditCredentials {
-    /// Reddit's OAuth docs call these "client id" and "client secret"; the app
-    /// registration UI often labels them "app id" / "app secret" — same values.
     fn from_env() -> Option<Self> {
         let client_id = std::env::var("REDDIT_CLIENT_ID")
             .or_else(|_| std::env::var("REDDIT_APP_ID"))
@@ -128,12 +152,10 @@ pub fn default_user_agent() -> String {
     })
 }
 
-/// True when this node can be loaded from the Reddit JSON API.
 pub fn is_fetchable(id: &ItemId) -> bool {
     !map_item_to_reddit_api(id, "https://example.com").is_empty()
 }
 
-/// Derive UI-facing fields from a stored payload (Reddit-specific when under reddit.com).
 pub fn entity_view_from_payload(id: &ItemId, payload: &Value) -> Option<crate::reducer::EntityData> {
     if id.as_str().starts_with("reddit.com") {
         return parse_reddit_view(id, payload);
@@ -141,12 +163,17 @@ pub fn entity_view_from_payload(id: &ItemId, payload: &Value) -> Option<crate::r
     None
 }
 
-/// Apply a full API payload to the in-memory tree (view derived for known domains).
 pub fn apply_entity_import(tree: &mut GlobalTree, id: &ItemId, payload: Value) {
     let view = entity_view_from_payload(id, &payload);
     tree.apply_entity_raw(id, payload, view);
 }
 
+fn notify(done: Option<oneshot::Sender<FetchJobResult>>, result: FetchJobResult) {
+    if let Some(tx) = done {
+        let _ = tx.send(result);
+    }
+}
+
 async fn reddit_worker(
     mut rx: mpsc::Receiver<RedditCommand>,
     tree: Arc<RwLock<GlobalTree>>,
@@ -168,15 +195,25 @@ async fn reddit_worker(
         recently_fetched.retain(|_, t| now.duration_since(*t) < cache_ttl);
 
         if in_flight.contains(&cmd.id) {
+            tracing::debug!(item = %cmd.id, "reddit fetch skipped: already in flight");
+            notify(cmd.done, FetchJobResult::SkippedDuplicate);
             continue;
         }
         if !cmd.force && recently_fetched.contains_key(&cmd.id) {
+            tracing::debug!(item = %cmd.id, "reddit fetch skipped: recently fetched cache");
+            notify(cmd.done, FetchJobResult::SkippedCached);
             continue;
         }
 
         in_flight.insert(cmd.id.clone());
         let fetch_id = cmd.id.clone();
+        let done = cmd.done;
 
+        tracing::debug!(
+            item = %fetch_id,
+            delay_ms = current_delay.as_millis(),
+            "reddit fetch starting after delay"
+        );
         tokio::time::sleep(current_delay).await;
 
         if let Some(c) = &creds {
@@ -184,8 +221,18 @@ async fn reddit_worker(
         }
 
         let token = oauth.as_ref().map(|t| t.access_token.as_str());
+        if token.is_some() {
+            tracing::debug!(item = %fetch_id, "reddit fetch using OAuth bearer");
+        }
+
+        let fetch_base = if token.is_some() {
+            &oauth_base
+        } else {
+            &api_base
+        };
+        let outcome = do_fetch(&client, fetch_base, &fetch_id, token).await;
 
-        match do_fetch(&client, &api_base, &fetch_id, token).await {
+        match outcome {
             Ok(FetchOutcome::Payload(payload)) => {
                 let ts = now_ms();
                 let event = Event::EntityImported {
@@ -193,30 +240,49 @@ async fn reddit_worker(
                     ts,
                     payload: payload.clone(),
                 };
-                if let Err(e) = event_log.append(&event).await {
-                    tracing::warn!("event log append failed for {}: {}", fetch_id, e);
-                } else {
-                    apply_entity_import(&mut *tree.write().await, &fetch_id, payload);
-                    recently_fetched.insert(fetch_id.clone(), Instant::now());
-                    current_delay = Duration::from_millis(600);
+                tracing::debug!(
+                    item = %fetch_id,
+                    ts,
+                    payload_keys = ?payload.as_object().map(|o| o.len()),
+                    "reddit fetch got JSON payload, appending event"
+                );
+                match event_log.append(&event).await {
+                    Err(e) => {
+                        tracing::warn!(
+                            item = %fetch_id,
+                            err = %e,
+                            "reddit event log append failed"
+                        );
+                        notify(done, FetchJobResult::Failed(e.to_string()));
+                    }
+                    Ok(()) => {
+                        apply_entity_import(&mut *tree.write().await, &fetch_id, payload);
+                        recently_fetched.insert(fetch_id.clone(), Instant::now());
+                        current_delay = Duration::from_millis(600);
+                        tracing::info!(item = %fetch_id, "reddit entity imported");
+                        notify(done, FetchJobResult::Imported);
+                    }
                 }
             }
             Ok(FetchOutcome::NotFound) => {
+                tracing::debug!(item = %fetch_id, "reddit fetch: not found (no event written)");
                 recently_fetched.insert(fetch_id.clone(), Instant::now());
+                notify(done, FetchJobResult::NotFound);
             }
             Ok(FetchOutcome::RateLimited { reset_secs }) => {
-                let wait = Duration::from_secs(reset_secs.max(1));
                 tracing::warn!(
-                    "Reddit rate limit for {}; sleeping {}s",
-                    fetch_id,
-                    wait.as_secs()
+                    item = %fetch_id,
+                    reset_secs,
+                    "reddit rate limited"
                 );
-                tokio::time::sleep(wait).await;
+                tokio::time::sleep(Duration::from_secs(reset_secs.max(1))).await;
                 current_delay = (current_delay * 2).min(Duration::from_secs(60));
+                notify(done, FetchJobResult::RateLimited { reset_secs });
             }
             Err(e) => {
-                tracing::warn!("Reddit fetch failed for {}: {}", fetch_id, e);
+                tracing::warn!(item = %fetch_id, err = %e, "reddit fetch failed");
                 current_delay = (current_delay * 2).min(Duration::from_secs(60));
+                notify(done, FetchJobResult::Failed(e));
             }
         }
 
@@ -238,6 +304,7 @@ async fn ensure_oauth_token(
 ) -> Option<OAuthToken> {
     if let Some(t) = existing {
         if Instant::now() < t.expires_at - Duration::from_secs(60) {
+            tracing::debug!("reddit OAuth token still valid");
             return Some(t);
         }
     }
@@ -246,6 +313,7 @@ async fn ensure_oauth_token(
         "{}/api/v1/access_token",
         oauth_base.trim_end_matches('/')
     );
+    tracing::debug!(%url, "reddit OAuth token request");
 
     let resp = client
         .post(&url)
@@ -257,13 +325,13 @@ async fn ensure_oauth_token(
     let resp = match resp {
         Ok(r) => r,
         Err(e) => {
-            tracing::warn!("Reddit OAuth token request failed: {e}");
+            tracing::warn!("reddit OAuth token request failed: {e}");
             return None;
         }
     };
 
     if !resp.status().is_success() {
-        tracing::warn!("Reddit OAuth token HTTP {}", resp.status());
+        tracing::warn!("reddit OAuth token HTTP {}", resp.status());
         return None;
     }
 
@@ -276,11 +344,12 @@ async fn ensure_oauth_token(
     let body: TokenResponse = match resp.json().await {
         Ok(b) => b,
         Err(e) => {
-            tracing::warn!("Reddit OAuth token parse failed: {e}");
+            tracing::warn!("reddit OAuth token parse failed: {e}");
             return None;
         }
     };
 
+    tracing::debug!(expires_in = body.expires_in, "reddit OAuth token acquired");
     Some(OAuthToken {
         access_token: body.access_token,
         expires_at: Instant::now() + Duration::from_secs(body.expires_in),
@@ -295,35 +364,72 @@ async fn do_fetch(
 ) -> Result<FetchOutcome, String> {
     let url = map_item_to_reddit_api(id, api_base);
     if url.is_empty() {
+        tracing::debug!(item = %id, "reddit do_fetch: no API URL for item");
         return Ok(FetchOutcome::NotFound);
     }
 
+    tracing::debug!(item = %id, %url, bearer = bearer.is_some(), "reddit HTTP GET");
+
     let mut req = client.get(&url);
     if let Some(token) = bearer {
         req = req.bearer_auth(token);
     }
 
-    let resp = req.send().await.map_err(|e| e.to_string())?;
+    let resp = req.send().await.map_err(|e| {
+        tracing::debug!(item = %id, %url, err = %e, "reddit HTTP transport error");
+        e.to_string()
+    })?;
+
+    let status = resp.status();
+    tracing::debug!(
+        item = %id,
+        %url,
+        %status,
+        remaining = ?rate_limit_remaining(&resp),
+        reset = ?rate_limit_reset_secs(&resp),
+        "reddit HTTP response"
+    );
 
-    if resp.status() == StatusCode::TOO_MANY_REQUESTS {
+    if status == StatusCode::TOO_MANY_REQUESTS {
         let reset = rate_limit_reset_secs(&resp);
         return Ok(FetchOutcome::RateLimited { reset_secs: reset });
     }
 
-    if resp.status() == StatusCode::SERVICE_UNAVAILABLE {
-        return Err("Reddit unavailable (503)".to_string());
+    if status == StatusCode::SERVICE_UNAVAILABLE {
+        return Err("Reddit unavailable (503)".into());
     }
 
-    if !resp.status().is_success() {
+    if !status.is_success() {
+        let body = resp.text().await.unwrap_or_default();
+        tracing::debug!(
+            item = %id,
+            %status,
+            body_len = body.len(),
+            body_prefix = %body.chars().take(240).collect::<String>(),
+            "reddit non-success body"
+        );
         return Ok(FetchOutcome::NotFound);
     }
 
     if rate_limit_remaining(&resp) == Some(0) {
         let reset = rate_limit_reset_secs(&resp);
+        tracing::debug!(item = %id, reset_secs = reset, "reddit headers: rate limit exhausted");
         return Ok(FetchOutcome::RateLimited { reset_secs: reset });
     }
 
-    let payload: Value = resp.json().await.map_err(|e| e.to_string())?;
+    let text = resp.text().await.map_err(|e| e.to_string())?;
+    tracing::debug!(item = %id, bytes = text.len(), "reddit response body received");
+
+    let payload: Value = serde_json::from_str(&text).map_err(|e| {
+        tracing::debug!(
+            item = %id,
+            err = %e,
+            body_prefix = %text.chars().take(240).collect::<String>(),
+            "reddit JSON parse failed"
+        );
+        format!("invalid JSON: {e}")
+    })?;
+
     Ok(FetchOutcome::Payload(payload))
 }
 
@@ -344,7 +450,6 @@ fn rate_limit_reset_secs(resp: &reqwest::Response) -> u64 {
         .unwrap_or(5)
 }
 
-/// Map canonical item id to a Reddit JSON API URL under `api_base`.
 pub fn map_item_to_reddit_api(id: &ItemId, api_base: &str) -> String {
     let path = id.as_str();
     if !path.starts_with("reddit.com/") && path != "reddit.com" {
@@ -445,26 +550,6 @@ mod tests {
             map_item_to_reddit_api(&id, "https://www.reddit.com"),
             "https://www.reddit.com/r/rust/about.json?raw_json=1"
         );
-        assert_eq!(
-            map_item_to_reddit_api(&id, "http://127.0.0.1:9999"),
-            "http://127.0.0.1:9999/r/rust/about.json?raw_json=1"
-        );
-    }
-
-    #[test]
-    fn map_post_url() {
-        let id = ItemId::parse("reddit.com/r/amitheasshole/comments/1trnvdl").unwrap();
-        assert_eq!(
-            map_item_to_reddit_api(&id, "https://www.reddit.com"),
-            "https://www.reddit.com/r/amitheasshole/comments/1trnvdl.json?raw_json=1"
-        );
-    }
-
-    #[test]
-    fn is_fetchable_reddit_sub() {
-        let id = ItemId::parse("reddit.com/r/rust").unwrap();
-        assert!(is_fetchable(&id));
-        assert!(!is_fetchable(&ItemId::opaque("example.com/x")));
     }
 
     #[test]
@@ -477,19 +562,5 @@ mod tests {
         )
         .unwrap();
         assert_eq!(entity.title, "The Rust Programming Language");
-        assert!(entity.body_html.as_ref().is_some_and(|b| b.contains("Rust")));
-    }
-
-    #[test]
-    fn parse_post_fixture() {
-        let json = r#"[{"kind":"Listing","data":{"children":[{"kind":"t3","data":{"title":"AITA","author":"op","selftext_html":"&lt;p&gt;hi&lt;/p&gt;","thumbnail":"https://b.thumbs.redditmedia.com/x.jpg"}}]}}]"#;
-        let v: Value = serde_json::from_str(json).unwrap();
-        let entity = entity_view_from_payload(
-            &ItemId::parse("reddit.com/r/x/comments/abc").unwrap(),
-            &v,
-        )
-        .unwrap();
-        assert_eq!(entity.title, "AITA");
-        assert_eq!(entity.author.as_deref(), Some("op"));
     }
 }
diff --git a/server/src/state.rs b/server/src/state.rs
index d71d1079a8f486cdab15795384aef0b81b32544d..e7ff9f663e45b5e168d6a3869948e3bd890966d4 100644
--- a/server/src/state.rs
+++ b/server/src/state.rs
@@ -143,9 +143,9 @@ impl AppState {
         Ok(())
     }
 
-    /// User-initiated Reddit/API import (via "Fetch more" — never on paste or navigate).
-    pub fn queue_entity_fetch(&self, id: ItemId) {
-        self.reddit.request_fetch(id, true);
+    /// User-initiated Reddit/API import (SSE / fetch module only).
+    pub fn queue_entity_fetch(&self, id: ItemId, done: Option<tokio::sync::oneshot::Sender<crate::reddit::FetchJobResult>>) {
+        self.reddit.request_fetch(id, true, done);
     }
 
     pub async fn record_vote(
diff --git a/server/src/ui_action.rs b/server/src/ui_action.rs
index 3d6a49a2a3fb950752827efe1f5a049308f09baf..ef9ac873fc0442e0b960f144309c8e91124d7884 100644
--- a/server/src/ui_action.rs
+++ b/server/src/ui_action.rs
@@ -26,7 +26,7 @@ pub enum HtmlUiAction {
     ParseQuery {
         query: String,
     },
-    /// Fetch upstream entity data for the current page (explicit user action only).
+    /// Import entity data; `POST /ui` responds with `text/event-stream` (not JS).
     FetchEntity {
         item: String,
     },
diff --git a/server/static/sorter_ui.js b/server/static/sorter_ui.js
index d9b016f197547f37ca4b2bcdd7ee6b673d0fb3f0..fe7bfc657c0fccae5c596d005b0e807abefec677 100644
--- a/server/static/sorter_ui.js
+++ b/server/static/sorter_ui.js
@@ -1,5 +1,5 @@
 /**
- * sorter2 web UI: fetch/eval for POST /ui. No product logic here.
+ * sorter2 web UI: POST /ui returns JS (morph) or SSE (entity fetch).
  */
 (function () {
   function evalJs(js) {
@@ -8,15 +8,98 @@
     }
   }
 
+  function morphSelector(selector, html) {
+    var el = document.querySelector(selector);
+    if (el && typeof Idiomorph !== 'undefined') {
+      Idiomorph.morph(el, html);
+    }
+  }
+
+  function handleSseEvent(eventType, data, form) {
+    if (eventType === 'fetching' || eventType === 'complete') {
+      try {
+        var msg = JSON.parse(data);
+        morphSelector(msg.selector || '#entity-section', msg.html);
+      } catch (err) {
+        console.warn('fetch morph parse', err);
+      }
+    }
+    if (eventType === 'complete' || eventType === 'error') {
+      var btn = form && form.querySelector('button[type="submit"]');
+      if (btn) btn.disabled = false;
+    }
+    if (eventType === 'error') {
+      try {
+        var err = JSON.parse(data);
+        console.warn('fetch error:', err.message || data);
+      } catch (_e) {
+        console.warn('fetch error:', data);
+      }
+    }
+  }
+
+  function consumeSseStream(response, form) {
+    var reader = response.body.getReader();
+    var decoder = new TextDecoder();
+    var buffer = '';
+    var eventType = '';
+    var dataLines = [];
+
+    function dispatch() {
+      if (!eventType && dataLines.length === 0) return;
+      handleSseEvent(eventType || 'message', dataLines.join('\n'), form);
+      eventType = '';
+      dataLines = [];
+    }
+
+    function pump() {
+      return reader.read().then(function (chunk) {
+        if (chunk.done) {
+          dispatch();
+          return;
+        }
+        buffer += decoder.decode(chunk.value, { stream: true });
+        var parts = buffer.split('\n');
+        buffer = parts.pop() || '';
+        for (var i = 0; i < parts.length; i++) {
+          var line = parts[i].replace(/\r$/, '');
+          if (line === '') {
+            dispatch();
+          } else if (line.indexOf('event:') === 0) {
+            eventType = line.slice(6).trim();
+          } else if (line.indexOf('data:') === 0) {
+            dataLines.push(line.slice(5).trim());
+          }
+        }
+        return pump();
+      });
+    }
+
+    return pump();
+  }
+
   function postUiForm(form) {
+    var btn = form.querySelector('button[type="submit"]');
+    if (form.id === 'fetch-entity-form' && btn) {
+      btn.disabled = true;
+    }
     return fetch(form.action, {
       method: 'POST',
       body: new URLSearchParams(new FormData(form)),
       headers: { 'Content-Type': 'application/x-www-form-urlencoded' },
       credentials: 'same-origin',
     }).then(function (resp) {
-      return resp.text();
-    }).then(evalJs);
+      var ct = resp.headers.get('content-type') || '';
+      if (ct.indexOf('text/event-stream') !== -1) {
+        return consumeSseStream(resp, form);
+      }
+      return resp.text().then(evalJs);
+    }).catch(function (err) {
+      if (form.id === 'fetch-entity-form' && btn) {
+        btn.disabled = false;
+      }
+      console.warn('POST /ui failed', err);
+    });
   }
 
   function initSorterUi() {
diff --git a/test/reddit_import.clj b/test/reddit_import.clj
index 54edaedeef08718c8f184d6aa468d8e6415068c7..17f2ee39e1498b0b6bb6a9d8a7e35608b0bd1b32 100644
--- a/test/reddit_import.clj
+++ b/test/reddit_import.clj
@@ -44,10 +44,12 @@
             (do (Thread/sleep 200) (recur))
             false))))))
 
-(defn- curl-post-ui [base rpc-json]
+(defn- curl-fetch-ui-sse [base item]
   (process/shell {:out :string :err :string}
-                 "curl" "-sf" "-X" "POST" (str base "/ui")
-                 "--data-urlencode" (str "__rpc__=" rpc-json)))
+                 "curl" "-sfN" "--max-time" "20"
+                 "-X" "POST" (str base "/ui")
+                 "--data-urlencode"
+                 (str "__rpc__={\"action\":\"fetch_entity\",\"item\":\"" item "\"}")))
 
 (defn- wait-event-log [path ms]
   (let [deadline (+ (System/currentTimeMillis) ms)]
@@ -98,11 +100,12 @@
                                               "curl" "-sf" browse-url))]
               (is (str/includes? before "Fetch from Reddit"))
               (is (not (str/includes? before "The Rust Programming Language")))
-              (let [rpc "{\"action\":\"fetch_entity\",\"item\":\"reddit.com/r/rust\"}"
-                    post (curl-post-ui app-base rpc)
-                    log-path (str data-dir "/events.jsonl")]
-                (is (zero? (:exit post)) "fetch_entity POST succeeds")
-                (is (wait-event-log log-path 10000) "event log written")
+              (let [log-path (str data-dir "/events.jsonl")
+                    sse (curl-fetch-ui-sse app-base "reddit.com/r/rust")]
+                (is (zero? (:exit sse)) "POST /ui fetch_entity SSE succeeds")
+                (is (str/includes? (:out sse) "event: complete"))
+                (is (str/includes? (:out sse) "The Rust Programming Language"))
+                (is (wait-event-log log-path 2000) "event log written")
                 (let [after (:out (process/shell {:out :string :err :string}
                                                  "curl" "-sf" browse-url))
                       log (slurp (io/file log-path))]