feat: add board goal context to agent templates

backend/app/services/agent_provisioning.py

@@ -129,6 +129,13 @@ def _build_context(
         "agent_name": agent.name,
         "agent_id": agent_id,
         "board_id": str(board.id),
+        "board_name": board.name,
+        "board_type": board.board_type,
+        "board_objective": board.objective or "",
+        "board_success_metrics": json.dumps(board.success_metrics or {}),
+        "board_target_date": board.target_date.isoformat() if board.target_date else "",
+        "board_goal_confirmed": str(board.goal_confirmed).lower(),
+        "is_board_lead": str(agent.is_board_lead).lower(),
         "session_key": session_key,
         "workspace_path": workspace_path,
         "base_url": base_url,

templates/AGENTS.md

@@ -29,6 +29,7 @@ Write things down. Do not rely on short-term context.
 
 ## Heartbeats
 - HEARTBEAT.md defines what to do on each heartbeat.
+- If **IS_BOARD_LEAD** is true, you are responsible for coordination and must run the Board Lead Loop.
 
 ## Task updates
 - All task updates MUST be posted to the task comments endpoint.

templates/HEARTBEAT.md

@@ -30,6 +30,26 @@ If any required input is missing, stop and request a provisioning update.
   - GET $BASE_URL/api/v1/boards/{BOARD_ID}/tasks must succeed.
 - If any check fails, stop and retry next heartbeat.
 
+## Board Lead Loop (if IS_BOARD_LEAD == true)
+When you are the board lead, run this loop after pre-flight checks and before claiming work:
+1) Read board goal context:
+   - Board: {{ board_name }} ({{ board_type }})
+   - Objective: {{ board_objective }}
+   - Success metrics: {{ board_success_metrics }}
+   - Target date: {{ board_target_date }}
+2) Review recent tasks/comments and board memory:
+   - GET $BASE_URL/api/v1/boards/{BOARD_ID}/tasks?limit=50
+   - GET $BASE_URL/api/v1/boards/{BOARD_ID}/memory?limit=50
+3) Update a short Board Plan Summary in board memory:
+   - POST $BASE_URL/api/v1/boards/{BOARD_ID}/memory
+     Body: {"content":"Plan summary + next gaps","tags":["plan","lead"],"source":"lead_heartbeat"}
+4) Identify missing steps and propose tasks.
+5) For each candidate task, compute confidence (rubric) and check risk/external actions:
+   - If risky/external or confidence < 80, create an approval:
+     - POST $BASE_URL/api/v1/boards/{BOARD_ID}/approvals
+   - Else create the task and assign an agent.
+6) Post a brief status update in board memory (1-3 bullets).
+
 ## Heartbeat checklist (run in order)
 1) Check in:
 ```bash

templates/USER.md

@@ -11,6 +11,14 @@
 
 {{ user_context }}
 
+## Board Goal
+
+- **Board name:** {{ board_name }}
+- **Board type:** {{ board_type }}
+- **Objective:** {{ board_objective }}
+- **Success metrics:** {{ board_success_metrics }}
+- **Target date:** {{ board_target_date }}
+
 ---
 
 The more you know, the better you can help. But remember -- you're learning about a person, not building a dossier. Respect the difference.