diff --git a/docs/coordinator.md b/docs/coordinator.md
index 09126c86..fadfd605 100644
--- a/docs/coordinator.md
+++ b/docs/coordinator.md
@@ -74,6 +74,33 @@ as it progresses through lifecycle phases (`"nix build"`, `"nixos-container stop
 `/api/state` and renders the current step beneath the running entry so the operator
 can see which phase is taking time.
 
+### Dependency tracking
+
+Each entry carries a `depends_on: Vec<u64>` field. The worker skips entries whose
+dependencies are not yet resolved — a dependency is resolved when its id is either
+in the queue as a terminal entry (`Done` / `Failed` / `Cancelled`) or no longer in
+the queue at all (evicted by `trim_history`, which only evicts terminals).
+
+Use cases:
+- Chain a `Rebuild` after an explicit prerequisite step without coupling them through
+  the `parent_id` cascade mechanism.
+- Sequence a `PermChange` + `Rebuild` pair where the rebuild must not start until the
+  perm-file write commits (already handled by the single-worker FIFO today, but
+  `depends_on` allows explicit cross-kind sequencing when parallel workers are added).
+
+`depends_on` is part of the dedup key: two entries with the same `(kind, agent,
+parent_id, inputs, approval_id)` but different dep sets are treated as distinct work.
+
+**Worker re-notification**: the worker drains `take_next()` in a tight loop after
+each entry finishes. When a dep entry transitions to terminal, the loop re-evaluates
+the queue immediately, so downstream entries are unblocked with no extra wakeup. No
+additional `notify_one()` call is needed.
+
+**Circular-dep caveat**: if A depends on B and B depends on A, neither entry ever
+becomes runnable — the worker skips both indefinitely with no error. Callers must
+ensure acyclic dep graphs. Cycle detection is deferred to a future iteration (when
+parallel workers make a stuck queue more visible).
+
 ---
 
 ## Container view
diff --git a/frontend/packages/dashboard/src/logs.js b/frontend/packages/dashboard/src/logs.js
index ab27fec9..4e9b4b4a 100644
--- a/frontend/packages/dashboard/src/logs.js
+++ b/frontend/packages/dashboard/src/logs.js
@@ -73,7 +73,7 @@ import {
       }
       const ul = el('ul', { class: 'build-logs-list' });
       for (const h of rows) {
-        const li = el('li', { class: 'build-logs-item' });
+        const li = el('li', { class: 'build-logs-item', 'data-log-id': String(h.id) });
         // status is null while running, 'ok'/'fail' when finished.
         const live = !h.status;
         const ok = h.status === 'ok';
@@ -175,6 +175,20 @@ import {
         ul.append(li);
       }
       buildList.append(ul);
+      // Deep-link: if the URL contains ?id=N, auto-expand that log entry
+      // so a `logs →` link from the rebuild-queue panel drops the operator
+      // straight into the live output without extra clicks.
+      const deepId = new URLSearchParams(location.search).get('id');
+      if (deepId) {
+        const target = ul.querySelector('[data-log-id="' + deepId + '"]');
+        if (target) {
+          const btn = target.querySelector('.build-logs-row-btn');
+          if (btn) {
+            btn.click();
+            target.scrollIntoView({ behavior: 'smooth', block: 'start' });
+          }
+        }
+      }
     } catch (err) {
       buildList.replaceChildren();
       buildList.append(el('p', { class: 'meta' }, 'fetch failed: ' + err));
diff --git a/frontend/packages/dashboard/src/tabs.js b/frontend/packages/dashboard/src/tabs.js
index d28e0568..9895687a 100644
--- a/frontend/packages/dashboard/src/tabs.js
+++ b/frontend/packages/dashboard/src/tabs.js
@@ -2266,6 +2266,20 @@ window.marked = marked;
     if (entry.step) {
       li.append(el('div', { class: 'rqe-step' }, '↳ ' + entry.step));
     }
+    // Live-log link: when a build_log_id is present the update/create op
+    // opened a build_logs row; the SSE stream endpoint lets the operator
+    // follow output in real time without polling.
+    if (entry.build_log_id != null) {
+      li.append(
+        ' ',
+        el('a', {
+          class: 'rqe-log-link',
+          href: '/logs.html?id=' + entry.build_log_id + '#build',
+          target: '_blank',
+          title: 'view build log #' + entry.build_log_id,
+        }, 'logs →'),
+      );
+    }
     // Error block, when failed.
     if (entry.error) {
       li.append(el('pre', { class: 'rqe-error', title: entry.error }, truncate(entry.error, 200)));
diff --git a/hive-c0re/src/actions.rs b/hive-c0re/src/actions.rs
index 2f70c4c8..9b573ec6 100644
--- a/hive-c0re/src/actions.rs
+++ b/hive-c0re/src/actions.rs
@@ -55,6 +55,7 @@ pub async fn approve(coord: Arc<Coordinator>, id: i64) -> Result<()> {
                 Vec::new(),
                 Some(id),
                 None,
+                Vec::new(),
             );
             coord.emit_rebuild_queue_snapshot();
             Ok(())
@@ -74,6 +75,7 @@ pub async fn approve(coord: Arc<Coordinator>, id: i64) -> Result<()> {
                 inputs.clone(),
                 Some(id),
                 None,
+                Vec::new(),
             );
             // Pre-enqueue cascade rebuilds in topological order so
             // agents depending on updated inputs are rebuilt after the
@@ -102,6 +104,7 @@ pub async fn approve(coord: Arc<Coordinator>, id: i64) -> Result<()> {
                 Vec::new(),
                 Some(id),
                 None,
+                Vec::new(),
             );
             coord.emit_rebuild_queue_snapshot();
             Ok(())
@@ -547,9 +550,19 @@ async fn run_apply_commit(
     // "nixos-container update" label for the whole multi-minute window.
     let hive = coord.hive_env();
     let paths = Coordinator::agent_paths(&approval.agent, agent_dir.to_path_buf());
-    let build_result = lifecycle::rebuild_no_meta(&approval.agent, &hive, &paths, &|step| {
-        coord.set_queue_step(queue_entry_id, step)
-    })
+    let build_result = lifecycle::rebuild_no_meta(
+        &approval.agent,
+        &hive,
+        &paths,
+        &|step| coord.set_queue_step(queue_entry_id, step),
+        &|log_id| {
+            if let Some(qid) = queue_entry_id {
+                if coord.rebuild_queue.set_build_log_id(qid, log_id) {
+                    coord.emit_rebuild_queue_snapshot();
+                }
+            }
+        },
+    )
     .await;
 
     match build_result {
diff --git a/hive-c0re/src/auto_update.rs b/hive-c0re/src/auto_update.rs
index 44886d86..3f21a210 100644
--- a/hive-c0re/src/auto_update.rs
+++ b/hive-c0re/src/auto_update.rs
@@ -84,9 +84,19 @@ pub async fn rebuild_agent(
     // lifecycle_action; this catches the auto-update scan + any
     // other direct caller.
     let guard = coord.transient_guard(name, crate::coordinator::TransientKind::Rebuilding);
-    let result = lifecycle::rebuild(name, &hive, &paths, &|step| {
-        coord.set_queue_step(queue_entry_id, step)
-    })
+    let result = lifecycle::rebuild(
+        name,
+        &hive,
+        &paths,
+        &|step| coord.set_queue_step(queue_entry_id, step),
+        &|log_id| {
+            if let Some(qid) = queue_entry_id {
+                if coord.rebuild_queue.set_build_log_id(qid, log_id) {
+                    coord.emit_rebuild_queue_snapshot();
+                }
+            }
+        },
+    )
     .await;
     drop(guard);
     match &result {
diff --git a/hive-c0re/src/lifecycle.rs b/hive-c0re/src/lifecycle.rs
index 0797409a..c11c3bc0 100644
--- a/hive-c0re/src/lifecycle.rs
+++ b/hive-c0re/src/lifecycle.rs
@@ -377,6 +377,7 @@ pub async fn rebuild(
     hive: &HiveEnv,
     paths: &AgentPaths,
     on_step: &(dyn Fn(&str) + Send + Sync),
+    on_build_log_id: &(dyn Fn(i64) + Send + Sync),
 ) -> Result<()> {
     // Sync the meta flake (idempotent — no-op when the rendered
     // flake matches disk) so a manual rebuild from the dashboard
@@ -389,7 +390,7 @@ pub async fn rebuild(
     // `applied/<n>/main` currently points at (deployed/<latest>).
     // Commits the lock if it changed.
     crate::meta::lock_update_for_rebuild(name).await?;
-    rebuild_no_meta(name, hive, paths, on_step).await
+    rebuild_no_meta(name, hive, paths, on_step, on_build_log_id).await
 }
 
 /// Container-level rebuild without touching the meta repo. Callers
@@ -402,11 +403,17 @@ pub async fn rebuild(
 /// label so callers can surface progress (e.g. update the rebuild-queue
 /// step shown in the dashboard). Pass `&|_| ()` when progress reporting
 /// is not needed.
+///
+/// `on_build_log_id` is called with the build-log row id immediately after
+/// the `nixos-container update` log row opens, before the actual update
+/// command starts. Callers can use this to link the queue entry to the log
+/// for live streaming. Pass `&|_| ()` when not needed.
 pub async fn rebuild_no_meta(
     name: &str,
     hive: &HiveEnv,
     paths: &AgentPaths,
     on_step: &(dyn Fn(&str) + Send + Sync),
+    on_build_log_id: &(dyn Fn(i64) + Send + Sync),
 ) -> Result<()> {
     validate(name)?;
     if let Some(other) = port_collision(name).await {
@@ -440,7 +447,7 @@ pub async fn rebuild_no_meta(
             priv_run("stop", name).await?;
         }
         on_step("nixos-container update");
-        let update_result = priv_run("update", name).await;
+        let update_result = priv_run_inner("update", name, Some(on_build_log_id)).await;
         if let Err(ref update_err) = update_result {
             // The update failed (e.g. nix build error). If the agent was
             // running before we stopped it, try to bring it back up on the
@@ -1322,6 +1329,24 @@ fn make_log_callback(
 /// is appended to the build-log row as it arrives, so the dashboard
 /// shows live progress during long `nixos-container create` / `update` runs.
 async fn priv_run(kind: &str, name: &str) -> Result<()> {
+    priv_run_inner(kind, name, None).await
+}
+
+/// Like `priv_run` but calls `on_log_id(log_id)` immediately after the
+/// build-log row is opened — before the actual container op starts.
+/// This lets callers surface the row id for live streaming (e.g. the
+/// rebuild-queue worker sets `build_log_id` on the queue entry so the
+/// dashboard can link to `/api/build-logs/id/{id}/stream`).
+///
+/// The callback fires only when a build-log row is successfully opened
+/// (i.e. the global `BuildLogs` handle is installed AND `h.start()`
+/// succeeds). No-op when `on_log_id` is `None` — that's the path for
+/// all callers that don't need the id.
+async fn priv_run_inner(
+    kind: &str,
+    name: &str,
+    on_log_id: Option<&(dyn Fn(i64) + Send + Sync)>,
+) -> Result<()> {
     let container = container_name(name);
     let cmdline = format!("nixos-container {kind} {container}");
 
@@ -1333,6 +1358,11 @@ async fn priv_run(kind: &str, name: &str) -> Result<()> {
             })
             .ok()
     });
+    // Notify the caller as soon as the log row exists so it can surface
+    // the id for live streaming before the container op even starts.
+    if let (Some(id), Some(cb)) = (log_id, on_log_id) {
+        cb(id);
+    }
 
     // For long-running ops use the streaming protocol so build_logs
     // receives lines in real time rather than as a batch at completion.
diff --git a/hive-c0re/src/rebuild_queue.rs b/hive-c0re/src/rebuild_queue.rs
index a5238454..e299b243 100644
--- a/hive-c0re/src/rebuild_queue.rs
+++ b/hive-c0re/src/rebuild_queue.rs
@@ -206,6 +206,23 @@ pub struct QueueEntry {
     /// the wire in those cases.
     #[serde(default, skip_serializing_if = "Option::is_none")]
     pub perm_payload: Option<PermPayload>,
+    /// Entries this entry must wait for before it can run. The worker
+    /// skips this entry until every id in the list has reached a
+    /// terminal state (`Done` / `Failed` / `Cancelled`) — or no longer
+    /// exists in the queue (evicted terminal entries are treated as
+    /// resolved, since `trim_history` only evicts terminals). Empty on
+    /// most entries; serialised only when non-empty.
+    #[serde(default, skip_serializing_if = "Vec::is_empty")]
+    pub depends_on: Vec<u64>,
+    /// Row id of the associated `build_logs` entry (opened by the
+    /// lifecycle worker when `nixos-container update` starts). Set
+    /// shortly after `state` transitions to `Running`; `None` while
+    /// `Queued` or for entries that don't open a build log (`Restart`,
+    /// `PermChange` file-write phase, etc.). Links the queue card to
+    /// the live-streaming `/api/build-logs/id/{id}/stream` endpoint so
+    /// the operator can follow the nix build output in real time.
+    #[serde(default, skip_serializing_if = "Option::is_none")]
+    pub build_log_id: Option<i64>,
 }
 
 /// How many terminal-state entries (`Done` / `Failed` / `Cancelled`)
@@ -286,6 +303,7 @@ impl RebuildQueue {
             Vec::new(),
             None,
             None,
+            Vec::new(),
         )
     }
 
@@ -303,7 +321,17 @@ impl RebuildQueue {
         parent_id: Option<u64>,
         inputs: Vec<String>,
     ) -> u64 {
-        self.enqueue_full(kind, agent, source, reason, parent_id, inputs, None, None)
+        self.enqueue_full(
+            kind,
+            agent,
+            source,
+            reason,
+            parent_id,
+            inputs,
+            None,
+            None,
+            Vec::new(),
+        )
     }
 
     /// Enqueue a `PermChange` entry for `agent`. The worker applies the
@@ -325,6 +353,7 @@ impl RebuildQueue {
             Vec::new(),
             None,
             Some(payload),
+            Vec::new(),
         )
     }
 
@@ -333,10 +362,10 @@ impl RebuildQueue {
     /// `enqueue_with_perm` delegate to this; the approval-driven POST
     /// handlers call it directly with the source row's id so the
     /// worker can re-fetch the kind-specific payload.
-    // 9 args: the queue entry has 6 independent submit-time fields plus
-    // three kind-specific payload fields (inputs, approval_id, perm_payload).
-    // A builder struct would obscure the call sites; the shorter wrappers
-    // already cover all common cases.
+    // 10 args: the queue entry has 6 independent submit-time fields plus
+    // four kind-specific payload fields (inputs, approval_id, perm_payload,
+    // depends_on). A builder struct would obscure the call sites; the
+    // shorter wrappers already cover all common cases.
     #[allow(clippy::too_many_arguments)]
     pub fn enqueue_full(
         &self,
@@ -348,6 +377,7 @@ impl RebuildQueue {
         inputs: Vec<String>,
         approval_id: Option<i64>,
         perm_payload: Option<PermPayload>,
+        depends_on: Vec<u64>,
     ) -> u64 {
         let mut inner = self.inner.lock().expect("rebuild_queue mutex poisoned");
         // Dedup against a pending entry with the same (kind, agent) —
@@ -380,6 +410,7 @@ impl RebuildQueue {
                 && entry.approval_id == approval_id
                 && entry.parent_id == parent_id
                 && perm_type_matches
+                && entry.depends_on == depends_on
             {
                 if !entry.reason.contains(&reason) {
                     use std::fmt::Write as _;
@@ -406,6 +437,8 @@ impl RebuildQueue {
             approval_id,
             step: None,
             perm_payload,
+            depends_on,
+            build_log_id: None,
         };
         inner.entries.push_back(entry);
         // Wake the worker. `notify_one` is a no-op when there's no
@@ -414,16 +447,47 @@ impl RebuildQueue {
         id
     }
 
-    /// Pop the next `Queued` entry and mark it `Running`. Returns the
-    /// entry (a clone — the original stays in the queue so live state
-    /// reflects "this is currently running"). Returns `None` when there's
-    /// nothing queued.
+    /// Pop the next `Queued` entry whose dependencies are resolved and
+    /// mark it `Running`. Returns the entry (a clone — the original
+    /// stays in the queue so live state reflects "this is currently
+    /// running"). Returns `None` when there's nothing queued OR every
+    /// queued entry has unresolved dependencies.
+    ///
+    /// A dependency is "resolved" when the dep's id is either:
+    /// - still in the queue AND in a terminal state (`Done` / `Failed`
+    ///   / `Cancelled`), OR
+    /// - no longer in the queue (evicted by `trim_history` — only
+    ///   terminal entries are ever evicted, so missing == completed).
     pub fn take_next(&self) -> Option<QueueEntry> {
         let mut inner = self.inner.lock().expect("rebuild_queue mutex poisoned");
-        let pos = inner
+        // Collect ids that are still in the queue and terminal. Entries
+        // absent from the queue are also considered resolved (see above).
+        let terminal_ids: std::collections::HashSet<u64> = inner
             .entries
             .iter()
-            .position(|e| e.state == QueueState::Queued)?;
+            .filter(|e| e.state.is_terminal())
+            .map(|e| e.id)
+            .collect();
+        // Active (non-terminal) ids: Queued + Running. Named `active_ids`
+        // rather than `queued_ids` because Running entries are included;
+        // used to distinguish "still in flight" from "evicted (= resolved)".
+        let active_ids: std::collections::HashSet<u64> = inner
+            .entries
+            .iter()
+            .filter(|e| !e.state.is_terminal())
+            .map(|e| e.id)
+            .collect();
+        let pos = inner.entries.iter().position(|e| {
+            e.state == QueueState::Queued
+                && e.depends_on.iter().all(|dep_id| {
+                    // Resolved if terminal in queue OR not in queue at all.
+                    // Note: circular deps (A depends on B, B depends on A)
+                    // silently deadlock — neither entry ever becomes runnable.
+                    // Not a problem in v1 (no callers yet), but callers must
+                    // ensure acyclic dep graphs.
+                    terminal_ids.contains(dep_id) || !active_ids.contains(dep_id)
+                })
+        })?;
         let entry = &mut inner.entries[pos];
         entry.state = QueueState::Running;
         entry.started_at = Some(now_unix());
@@ -472,6 +536,24 @@ impl RebuildQueue {
         true
     }
 
+    /// Link a `build_logs` row to a `Running` entry. Called by the
+    /// lifecycle worker when `nixos-container update` opens a build log
+    /// row so the dashboard can surface a "view logs" link while the
+    /// build is in flight. Returns `true` when the row was found and
+    /// the id was stored; `false` when the entry is no longer in the
+    /// queue or is not `Running`.
+    pub fn set_build_log_id(&self, id: u64, log_id: i64) -> bool {
+        let mut inner = self.inner.lock().expect("rebuild_queue mutex poisoned");
+        let Some(entry) = inner.entries.iter_mut().find(|e| e.id == id) else {
+            return false;
+        };
+        if entry.state != QueueState::Running {
+            return false;
+        }
+        entry.build_log_id = Some(log_id);
+        true
+    }
+
     /// Snapshot the queue for `/api/state` and `RebuildQueueChanged`.
     /// Cheap clone — entries are small (~hundreds of bytes each).
     pub fn snapshot(&self) -> Vec<QueueEntry> {
@@ -1131,6 +1213,8 @@ mod tests {
             None,
             Vec::new(),
             Some(42),
+            None,
+            Vec::new(),
         );
         let snap = q.snapshot();
         let entry = snap.iter().find(|e| e.id == id).expect("entry present");
@@ -1153,6 +1237,8 @@ mod tests {
             None,
             Vec::new(),
             Some(1),
+            None,
+            Vec::new(),
         );
         let b = q.enqueue_full(
             QueueKind::Rebuild,
@@ -1162,6 +1248,8 @@ mod tests {
             None,
             Vec::new(),
             Some(2),
+            None,
+            Vec::new(),
         );
         assert_ne!(a, b);
         assert_eq!(q.snapshot().len(), 2);
@@ -1175,6 +1263,8 @@ mod tests {
             None,
             Vec::new(),
             Some(1),
+            None,
+            Vec::new(),
         );
         assert_eq!(a, c);
         assert_eq!(q.snapshot().len(), 2);
@@ -1346,4 +1436,187 @@ mod tests {
             None
         );
     }
+
+    // --- depends_on tests ---
+
+    /// An entry whose dep is not yet terminal must be skipped by
+    /// `take_next`; it runs only after the dep finishes.
+    #[test]
+    fn depends_on_blocks_until_dep_is_terminal() {
+        let q = RebuildQueue::new();
+        let a = q.enqueue(
+            QueueKind::Rebuild,
+            "agent-a".to_owned(),
+            QueueSource::Manual,
+            "first".to_owned(),
+            None,
+        );
+        let b = q.enqueue_full(
+            QueueKind::Rebuild,
+            "agent-b".to_owned(),
+            QueueSource::Manual,
+            "second (blocked on a)".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![a],
+        );
+        // B depends on A — take_next should give A first.
+        let first = q.take_next().expect("a is ready");
+        assert_eq!(first.id, a);
+        // A is Running, not terminal — B must still be blocked.
+        assert!(q.take_next().is_none(), "b must be blocked while a runs");
+        // Finish A → B should now be available.
+        q.finish(a, QueueState::Done, None);
+        let second = q.take_next().expect("b unblocked after a done");
+        assert_eq!(second.id, b);
+    }
+
+    /// An entry whose dep finished and was evicted from history is
+    /// treated as resolved (eviction only happens to terminal entries).
+    #[test]
+    fn depends_on_evicted_dep_counts_as_resolved() {
+        let q = RebuildQueue::new();
+        // Fill the history cap for Rebuild so old terminals get evicted.
+        for i in 0..MAX_HISTORY_PER_KIND {
+            let id = q.enqueue(
+                QueueKind::Rebuild,
+                format!("filler-{i}"),
+                QueueSource::Manual,
+                "filler".to_owned(),
+                None,
+            );
+            q.take_next();
+            q.finish(id, QueueState::Done, None);
+        }
+        // `dep` gets enqueued, run, finished, and evicted by the
+        // next history-trimming call.
+        let dep = q.enqueue(
+            QueueKind::Rebuild,
+            "dep-agent".to_owned(),
+            QueueSource::Manual,
+            "dep".to_owned(),
+            None,
+        );
+        q.take_next();
+        // One more terminal to push `dep` out of the history window.
+        q.finish(dep, QueueState::Done, None);
+        let extra = q.enqueue(
+            QueueKind::Rebuild,
+            "extra".to_owned(),
+            QueueSource::Manual,
+            "extra".to_owned(),
+            None,
+        );
+        q.take_next();
+        q.finish(extra, QueueState::Done, None);
+        // `dep` should now be evicted.
+        assert!(
+            q.snapshot().iter().all(|e| e.id != dep),
+            "dep must be evicted from history"
+        );
+        // An entry that depends on the (evicted) dep must be immediately runnable.
+        let downstream = q.enqueue_full(
+            QueueKind::Rebuild,
+            "downstream".to_owned(),
+            QueueSource::Manual,
+            "downstream (dep evicted = resolved)".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![dep],
+        );
+        let got = q.take_next().expect("downstream runnable when dep evicted");
+        assert_eq!(got.id, downstream);
+    }
+
+    /// Dedup respects `depends_on`: two otherwise-identical entries with
+    /// different dep sets are distinct and must NOT collapse.
+    #[test]
+    fn depends_on_is_part_of_dedup_key() {
+        let q = RebuildQueue::new();
+        let dep1 = q.enqueue(
+            QueueKind::Rebuild,
+            "dep1".to_owned(),
+            QueueSource::Manual,
+            "d1".to_owned(),
+            None,
+        );
+        let dep2 = q.enqueue(
+            QueueKind::Rebuild,
+            "dep2".to_owned(),
+            QueueSource::Manual,
+            "d2".to_owned(),
+            None,
+        );
+        let a = q.enqueue_full(
+            QueueKind::Rebuild,
+            "target".to_owned(),
+            QueueSource::Manual,
+            "r".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![dep1],
+        );
+        // Same kind+agent but different depends_on — must NOT dedup.
+        let b = q.enqueue_full(
+            QueueKind::Rebuild,
+            "target".to_owned(),
+            QueueSource::Manual,
+            "r".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![dep2],
+        );
+        assert_ne!(a, b, "different depends_on must produce distinct entries");
+        // Same depends_on as a — must dedup.
+        let c = q.enqueue_full(
+            QueueKind::Rebuild,
+            "target".to_owned(),
+            QueueSource::Manual,
+            "r again".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![dep1],
+        );
+        assert_eq!(a, c, "identical depends_on must dedup");
+    }
+
+    /// An entry with Failed dep is still resolved — the dependent runs
+    /// regardless of whether its upstream succeeded or not. Callers that
+    /// need to abort on dep failure should cancel the downstream manually.
+    #[test]
+    fn depends_on_failed_dep_counts_as_resolved() {
+        let q = RebuildQueue::new();
+        let a = q.enqueue(
+            QueueKind::Rebuild,
+            "a".to_owned(),
+            QueueSource::Manual,
+            "a".to_owned(),
+            None,
+        );
+        let b = q.enqueue_full(
+            QueueKind::Rebuild,
+            "b".to_owned(),
+            QueueSource::Manual,
+            "b (blocked on a)".to_owned(),
+            None,
+            Vec::new(),
+            None,
+            None,
+            vec![a],
+        );
+        q.take_next(); // pop a, mark Running
+        q.finish(a, QueueState::Failed, Some("nix build exploded".to_owned()));
+        let got = q.take_next().expect("b runnable after a failed");
+        assert_eq!(got.id, b);
+    }
 }
diff --git a/hive-c0re/src/server.rs b/hive-c0re/src/server.rs
index 12cacc92..0fc79567 100644
--- a/hive-c0re/src/server.rs
+++ b/hive-c0re/src/server.rs
@@ -161,7 +161,7 @@ async fn dispatch(req: &HostRequest, coord: Arc<Coordinator>) -> HostResponse {
                 let agent_dir = coord.ensure_runtime(name)?;
                 let hive = coord.hive_env();
                 let paths = Coordinator::agent_paths(name, agent_dir);
-                let result = lifecycle::rebuild(name, &hive, &paths, &|_| ()).await;
+                let result = lifecycle::rebuild(name, &hive, &paths, &|_| (), &|_| ()).await;
                 // Mirror auto_update::rebuild_agent — the manager wants
                 // to know about every rebuild attempt regardless of
                 // which surface triggered it, especially failures