opencastle 0.31.7 → 0.32.0

package/src/orchestrator/prompts/assess-complexity.prompt.md

@@ -43,6 +43,14 @@ Analyze the PRD below and produce a complexity assessment as a **single JSON obj
       "phases": [<phase numbers>],
       "depends_on": ["<group name>", ...]
     }
+  ],
+  "task_complexity": [
+    {
+      "workstream": "<workstream title from PRD Task Breakdown>",
+      "phase": <phase number>,
+      "complexity": 1 | 2 | 3 | 5 | 8 | 13,
+      "rationale": "<brief reason>"
+    }
   ]
 }
 ```
@@ -65,3 +73,8 @@ Analyze the PRD below and produce a complexity assessment as a **single JSON obj
   - **Minimum 3 tasks per group.** Never create a group that would produce a convoy with only 1–2 tasks — merge small groups with adjacent ones. A convoy with a single task is pointless overhead.
   - **Do NOT map phases 1:1 to groups.** Groups should bundle multiple related phases when tasks are tightly coupled (e.g., config + data in one group, components + pages in another). Only split at genuine domain boundaries where failure isolation matters.
   - Maximum 3 groups for projects with ≤ 15 tasks. Maximum 4 groups for 16+ tasks.
+- `task_complexity`: Array of per-workstream complexity assessments using Fibonacci scale.
+  - `workstream`: Exact workstream title from the PRD's Task Breakdown section.
+  - `phase`: Phase number the workstream belongs to.
+  - `complexity`: Fibonacci score: `1` (trivial — single file edit), `2` (simple — small fix, one test), `3` (moderate — new component/endpoint), `5` (significant — multi-file feature), `8` (complex — cross-cutting), `13` (epic — architecture-level).
+  - `rationale`: One sentence explaining the score.

package/src/orchestrator/prompts/generate-convoy.prompt.md

@@ -52,6 +52,13 @@ You are the Team Lead. The user wants to run `opencastle run` to execute a batch
 | `max_retries` | integer | no | Override retry count for this task |
 | `review` | `fast` \| `panel` \| `none` \| `auto` | no | Review level |
 | `gates` | list of strings | no | Per-task gate commands |
+| `complexity` | `1` \| `2` \| `3` \| `5` \| `8` \| `13` | no | Fibonacci complexity score. Auto-populates timeout/retries/review from effort table. |
+
+### Complexity Scores
+
+If a **Pre-Computed Task Complexity** table is present in the context, use those scores directly — do not re-derive them. Map each convoy task to the closest workstream; for combined workstreams use the highest score.
+
+If no pre-computed data is available, assess complexity yourself: `1` (trivial), `2` (simple), `3` (moderate), `5` (significant), `8` (complex), `13` (epic).
 
 ### Agent Roster
 
@@ -226,3 +233,15 @@ When chain mode is detected:
 }
 ````
 
+## Historical Performance Context
+
+When historical execution data is available (via `opencastle insights --json`), the Team Lead should include a compact summary in the context. Example:
+
+### Historical Performance (auto-generated)
+Based on {N} past convoys:
+- Tasks on `src/cli/` succeed 92% with Developer (avg 8min)
+- Recommended concurrency for {task_count} tasks: {recommended}
+- Agents to watch: {high_retry_agents}
+
+Use this data to make evidence-based decisions about agent assignment, concurrency, and timeout settings.
+