huskies: merge 543_story_resume_failed_coder_agents_with_resume_instead_of_starting_fresh_sessions

server/src/agents/pool/pipeline/advance.rs

@@ -17,6 +17,7 @@ use super::super::{AgentPool, StoryAgent};
 impl AgentPool {
     /// Pipeline advancement: after an agent completes, move the story to
     /// the next pipeline stage and start the appropriate agent.
+    #[allow(clippy::too_many_arguments)]
     pub(super) async fn run_pipeline_advance(
         &self,
         story_id: &str,
@@ -25,6 +26,7 @@ impl AgentPool {
         project_root: Option<PathBuf>,
         worktree_path: Option<PathBuf>,
         merge_failure_reported: bool,
+        previous_session_id: Option<String>,
     ) {
         let project_root = match project_root {
             Some(p) => p,
@@ -81,7 +83,7 @@ impl AgentPool {
                             if let Err(e) = crate::agents::lifecycle::move_story_to_qa(&project_root, story_id) {
                                 slog_error!("[pipeline] Failed to move '{story_id}' to 3_qa/: {e}");
                             } else if let Err(e) = self
-                                .start_agent(&project_root, story_id, Some("qa"), None)
+                                .start_agent(&project_root, story_id, Some("qa"), None, None)
                                 .await
                             {
                                 slog_error!("[pipeline] Failed to start qa agent for '{story_id}': {e}");
@@ -118,7 +120,13 @@ impl AgentPool {
                             completion.gate_output
                         );
                         if let Err(e) = self
-                            .start_agent(&project_root, story_id, Some(agent_name), Some(&context))
+                            .start_agent(
+                                &project_root,
+                                story_id,
+                                Some(agent_name),
+                                Some(&context),
+                                previous_session_id,
+                            )
                             .await
                         {
                             slog_error!(
@@ -202,7 +210,7 @@ impl AgentPool {
                             coverage_output
                         );
                         if let Err(e) = self
-                            .start_agent(&project_root, story_id, Some("qa"), Some(&context))
+                            .start_agent(&project_root, story_id, Some("qa"), Some(&context), None)
                             .await
                         {
                             slog_error!("[pipeline] Failed to restart qa for '{story_id}': {e}");
@@ -223,7 +231,7 @@ impl AgentPool {
                         completion.gate_output
                     );
                     if let Err(e) = self
-                        .start_agent(&project_root, story_id, Some("qa"), Some(&context))
+                        .start_agent(&project_root, story_id, Some("qa"), Some(&context), None)
                         .await
                     {
                         slog_error!("[pipeline] Failed to restart qa for '{story_id}': {e}");
@@ -322,6 +330,7 @@ impl AgentPool {
                                 story_id,
                                 Some("mergemaster"),
                                 Some(&context),
+                                None,
                             )
                             .await
                         {
@@ -369,7 +378,7 @@ impl AgentPool {
             return;
         }
         if let Err(e) = self
-            .start_agent(project_root, story_id, Some("mergemaster"), None)
+            .start_agent(project_root, story_id, Some("mergemaster"), None, None)
             .await
         {
             slog_error!("[pipeline] Failed to start mergemaster for '{story_id}': {e}");
@@ -392,6 +401,7 @@ pub(super) fn spawn_pipeline_advance(
     worktree_path: Option<PathBuf>,
     watcher_tx: broadcast::Sender<WatcherEvent>,
     merge_failure_reported: bool,
+    previous_session_id: Option<String>,
 ) {
     let sid = story_id.to_string();
     let aname = agent_name.to_string();
@@ -410,6 +420,7 @@ pub(super) fn spawn_pipeline_advance(
             project_root,
             worktree_path,
             merge_failure_reported,
+            previous_session_id,
         )
         .await;
     });
@@ -524,6 +535,7 @@ mod tests {
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -565,6 +577,7 @@ mod tests {
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -606,6 +619,7 @@ mod tests {
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -639,6 +653,7 @@ mod tests {
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -706,6 +721,7 @@ stage = "qa"
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -777,6 +793,7 @@ stage = "qa"
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -883,6 +900,7 @@ stage = "qa"
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 
@@ -950,6 +968,7 @@ stage = "qa"
             Some(root.to_path_buf()),
             None,
             false,
+            None,
         )
         .await;
 

server/src/agents/pool/pipeline/completion.rs

@@ -83,6 +83,7 @@ impl AgentPool {
             project_root_for_advance,
             wt_path_for_advance,
             merge_failure_reported_for_advance,
+            session_id_for_advance,
         ) = {
             let mut agents = self.agents.lock().map_err(|e| e.to_string())?;
             let agent = agents.get_mut(&key).ok_or_else(|| {
@@ -94,8 +95,9 @@ impl AgentPool {
             let pr = agent.project_root.clone();
             let wt = agent.worktree_info.as_ref().map(|w| w.path.clone());
             let mfr = agent.merge_failure_reported;
+            let sid_advance = agent.session_id.clone();
             agents.remove(&key);
-            (tx, sid, pr, wt, mfr)
+            (tx, sid, pr, wt, mfr, sid_advance)
         };
 
         // Emit Done so wait_for_agent unblocks.
@@ -128,6 +130,7 @@ impl AgentPool {
                     project_root_for_advance,
                     wt_path_for_advance,
                     merge_failure_reported_for_advance,
+                    session_id_for_advance,
                 )
                 .await;
         });
@@ -277,6 +280,8 @@ pub(in crate::agents::pool) async fn run_server_owned_completion(
         lock.remove(&key);
         (tx, pr, wt, mfr)
     };
+    // The completed session's ID is used to resume if gates fail.
+    let previous_session_id = session_id.clone();
 
     // Emit Done so wait_for_agent unblocks.
     let _ = tx.send(AgentEvent::Done {
@@ -299,6 +304,7 @@ pub(in crate::agents::pool) async fn run_server_owned_completion(
         wt_path_for_advance,
         watcher_tx,
         merge_failure_reported_for_advance,
+        previous_session_id,
     );
 }