Skip to content

AGENT_CHARTER.md

Latest commit

 

History

History
40 lines (31 loc) · 1.75 KB

AGENT_CHARTER.md

File metadata and controls

40 lines (31 loc) · 1.75 KB

AGENT CHARTER

Version: 1.0.0 Status: DRAFT — This describes a proposed 6-agent role architecture that has not been implemented. The current system uses a single generic writer identity per participant. Tracked by task:doc-agent-charter. Related: AGENT_PROTOCOL.md is the canonical spec for the actual agent-native CLI and action kernel. This charter is about role decomposition, not the concrete command surface. Enforcement: HARD BOUNDARY VIOLATION = IMMEDIATE REJECT

Agent Roster & Scopes

1. Parser Agent (Ingest → Normalize)

  • Input: Raw Markdown/YAML strings or file refs
  • Output: Array of raw Task v1.0 objects
  • Forbidden: State mutation, adding estimates, rewriting stories

2. Planner Agent (Classify → Merge)

  • Input: Normalized Tasks
  • Output: Tasks with campaign + tentative depends-on edges
  • Forbidden: Set estimates, create new milestones

3. Graph Agent (Rebalance → Schedule)

  • Input: Merged plan graph
  • Output: Annotated DAG + Scheduled Artifacts
  • Forbidden: Change stories, priorities, or criteria; emit PlanPatch directly

4. QA Agent (Review Gate)

  • Input: Emitted PlanPatch
  • Output: Review findings (MUST/SHOULD/COULD checklist)
  • Forbidden: Modify the patch, suppress MUST violations

5. Coordinator Agent (Triage, Janitor, Rebalancer)

  • Input: Graph frontier changes + heartbeat stream
  • Forbidden: Claim tasks for itself, execute work

6. Worker Agent (Specialized execution)

  • Identity: agent:<skill>-<uuid>
  • MUST: Use optimistic claim → verify ownership post-materialize → perform work
  • Forbidden: Touch any task they do not own, alter estimates

Agent Role Pipeline

Agent role pipeline