菜单
Subagent registration protocol. Use when assigned a task UUID by an orchestrator — claim the task, pull context from engram, store all findings, and report results back.
用 Codex 或 Claude 帮你安装 复制这段 Prompt，粘贴到 Codex、Claude 或其他助手里，让它检查 Skill 页面并帮你完成安装。
基于 SOC 职业分类
Document architectural decisions as engraph ADRs. Store context, options considered, rationale, and consequences for queryable decision history.
Every commit must reference a valid engram task UUID. Covers format, validation, and why --no-verify is prohibited.
Orchestrator agent loop. Coordinates work across subagents using engram as shared state. Search before acting, dispatch by UUID, validate before finishing.
Plan feature implementation by creating an engram task hierarchy. Not a markdown plan — a queryable task tree any agent can retrieve.
Universal session end protocol. Run at the end of every agent session — closes open tasks, generates a handoff summary, syncs to remote, and validates state.
Universal session start protocol. Run at the beginning of every agent session — syncs from remote, opens a session, loads prior context, and surfaces the next action.
| name | engram-subagent-register |
| description | Subagent registration protocol. Use when assigned a task UUID by an orchestrator — claim the task, pull context from engram, store all findings, and report results back. |
You are a subagent. You received a task UUID. Your input comes from engram. Your output goes to engram.
Per-turn write rule: After every
bash,read,edit, orwritecall, store the finding immediately withengram_context_createorengram_reasoning_create+engram_relationship_create. Do not batch. Do not wait until the end. Every turn with substantive work must have an engram write.
engram_task_show id=<TASK_UUID>
engram_task_update id=<TASK_UUID> status="in_progress"
engram_relationship_connected entity_id=<TASK_UUID> max_depth=2
engram_ask query="<keywords from task title>"
Never start work based only on the title. Always check linked context.
For each finding as you discover it:
engram_context_create title="<finding title>" content="<what you found>" source="<file/command>"
engram_relationship_create source_id=<TASK_UUID> source_type=task target_id=<CTX_UUID> target_type=context relationship_type=relates_to agent="<name>"
For reasoning:
engram_reasoning_create title="<reasoning title>" task_id=<TASK_UUID> content="<interpretation>"
engram_relationship_create source_id=<TASK_UUID> source_type=task target_id=<RSN_UUID> target_type=reasoning relationship_type=explains agent="<name>"
engram_reasoning_create title="Completion report: <task title>" task_id=<TASK_UUID> content="## Result\n<outcome>\n\n## What I did\n- ...\n\n## Findings\n- ...\n\n## Status\nCOMPLETED|BLOCKED"
engram_relationship_create source_id=<TASK_UUID> source_type=task target_id=<REPORT_UUID> target_type=reasoning relationship_type=explains agent="<name>"
engram_task_update id=<TASK_UUID> status="done" outcome="<one-line summary>"
engram_task_update id=<TASK_UUID> status="blocked" reason="<why blocked, what next agent needs>"