一键导入
work
Use when routing tracked work through lifecycle lanes, resolving workstream context, or deciding which lifecycle skill owns the next entity transition.
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。
菜单
Use when routing tracked work through lifecycle lanes, resolving workstream context, or deciding which lifecycle skill owns the next entity transition.
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。
Routes adaptive multi-agent deliberation with fractal context cycles. Use when using /cdo, think/deep/debug/parliament work, or long runs paired with autoresearch scheduling.
Use when you need an evidence-first provenance report from `cass`, repo docs, `.lev`, `~/.agents/diagrams`, `qmd`, or Grep/Glob tools.
Use the codebase knowledge graph for structural code queries. Triggers on: explore the codebase, understand the architecture, what functions exist, show me the structure, who calls this function, what does X call, trace the call chain, find callers of, show dependencies, impact analysis, dead code, unused functions, high fan-out, refactor candidates, code quality audit, graph query syntax, Cypher query examples, edge types, how to use search_graph.
Handle blocked destructive commands. Use when dcg blocks rm -rf, git reset --hard, DROP DATABASE, kubectl delete, or when configuring agent safety guardrails.
Use when running a visible Lev autopilot that coordinates research, interview, arch, propose, AgentPing, warm-worker intent, realtime dashboard intent, and lev exec updates across the lifecycle.
Heartbeat-driven autonomous development loop using lev loop autodev, with interval, budget, and tick controls for in-process SDLC execution.
基于 SOC 职业分类
| name | work |
| description | Use when routing tracked work through lifecycle lanes, resolving workstream context, or deciding which lifecycle skill owns the next entity transition. |
| skill_type | router |
| category | lifecycle |
| output_template | hud |
| sub_skills | ["ws","capture","prior-art","propose","exec","close","handoff"] |
/work is the thin lifecycle spine. It does not re-implement capture,
proposal, execution, close, or handoff protocols. It resolves workstream
context, identifies the entity movement, and routes to the owning skill.
Lifecycle lane: Router
Entity movement: unknown -> routed
Workstream: resolve active workstream before writes or dispatch
Upstream: any lifecycle skill or user request
Downstream: /ws, /capture, /prior-art, /propose, /exec, /close, /handoff
Router: /work
HUD: end with 🧬 {ws} ⚡{exec_count} 📥{capture_count} ⏸️{paused_count} ✅{done_count} | 🚦{gate}={score} | ⏭️ {next} | 🔁{loop_state}
End lifecycle-skill responses with this inline HUD line:
🧬 {ws} ⚡{exec_count} 📥{capture_count} ⏸️{paused_count} ✅{done_count} | 🚦{gate}={score} | ⏭️ {next} | 🔁{loop_state}
The shared graph-footer partial is an authoring reference only until the Lev skill-builder pipeline can project FlowMind fragments into skill bodies. Do not make skills load or include that file at runtime.
| Lane | Entity movement | Owns |
|---|---|---|
| Shape | `memory -> captured | blocked` |
| Plan | `captured | aligned -> proposed |
| Exec | `execution_ready -> executing -> verified | blocked |
| Close | `verified -> closed | monitoring |
| Router | unknown -> routed | /work, /ws |
## Work Link section.QA, Pentagon, UltraQA, and ai-slop-cleaner are lifecycle gate overlays, not
separate lanes. /work routes by missing gate state; lane skills own the
details.
| Gate state | Meaning | Route |
|---|---|---|
needs_proof_design | Highest-risk claim, fail-closed acceptance, or owner-local test placement is unclear | /interview or /propose |
needs_proof_gates | A non-trivial task lacks execution.yaml.proof_gates | /propose |
needs_runtime_qa | Declared baseline, Pentagon, UltraQA, or harness runtime checks still need execution | /exec |
needs_quality_review | Cleanup, refactor, fallback, boundary, or AI-slop risk needs ai-slop-cleaner review | /exec |
needs_close_verdict | Work is implemented but proof-gate verdicts or residual risks are not sealed | /close |
proof_blocked | A proof gate fails or cannot run safely | /propose for contract repair or /capture for follow-up |
steps:
- id: resolve_workstream
action: Identify the active workstream or route to /ws.
validation: "A workstream id is known, or /ws is the next skill."
on_failure: "Do not write. Route to /ws find|resume or create a workstream."
- id: classify_entity
action: Name the entity, current state, target state, missing gate, and proof_gate_state.
validation: "Entity has path/id/uri plus current_state, target_state, and proof_gate_state when non-trivial."
on_failure: "Route to /capture if it is only in conversation memory."
- id: route_lane
action: Select the lifecycle lane owner.
validation: "Exactly one next skill is chosen from the lane table."
on_failure: "Ask one routing question or route to /prior-art for evidence."
- id: enforce_work_link
action: Ensure the next skill declares lane, movement, upstream, downstream, and HUD.
validation: "The selected skill has a Work Link or this session is updating it."
on_failure: "Use skill-builder semantics to patch the skill before relying on it."
- id: emit_hud
action: Report current workstream, gate, next entity, and loop state.
validation: "Response ends with the inline HUD line."
on_failure: "Add the HUD before ending."
| Situation | Route |
|---|---|
| User dumps ideas or a thread has unfiled content | /capture or /dump as capture --deep |
| Need evidence, provenance, lineage, or duplicate detection | /prior-art |
| Idea/design is aligned but not execution-ready | /propose |
Task has dna.yaml and execution.yaml with a verifier | /exec |
| Task needs proof design, proof gates, runtime QA, or quality review | route by proof_gate_state |
| Work is verified and needs sealing, learning, commit, or next recommendation | /close |
| Session is ending, compacting, or needs a resume prompt | /handoff |
| Workstream identity is missing, stale, split, or tangled | /ws |
Each lifecycle skill carries one block in this shape:
## <Work Link>
Lifecycle lane: {Shape|Plan|Exec|Close|Router}
Entity movement: `{from_state} -> {to_state}`
Workstream: resolve active workstream before writes or dispatch
Upstream: `{skill_or_none}`
Downstream: `{skill_or_none}`
Router: `/work`
HUD: end with `🧬 {ws} ⚡{exec_count} 📥{capture_count} ⏸️{paused_count} ✅{done_count} | 🚦{gate}={score} | ⏭️ {next} | 🔁{loop_state}`
Templates are authoring references. Copy or render them through the owning lane; do not make another skill depend on reading them at runtime.
| Template | Destination | Owner |
|---|---|---|
templates/report.md | .lev/pm/reports/ | /prior-art, /capture |
templates/plan.md | .lev/pm/plans/ | /propose |
templates/design.md | .lev/pm/designs/ | /interview, /propose |
templates/proposal.md | .lev/pm/proposals/ | /propose |
templates/spec.md | .lev/pm/specs/ | /propose |
templates/decision.md | .lev/pm/decisions/ | /close |
templates/validation-report.md | .lev/pm/validation-reports/ | /exec, /close |
| Excuse | Reality |
|---|---|
| "Work owns everything." | Work routes; lane skills own detailed protocols. |
| "A handoff markdown is enough." | Workstream YAML is durable identity; markdown is projection. |
| "The user asked to execute." | Execution still needs an execution-ready entity and verifier. |
| "Footer file will be loaded." | HUD is inline until FlowMind skill projection exists. |
| "I know the workstream." | Name it in the HUD or route to /ws. |