Skip to content
Agent Orchestration
Skill

/project-life

Единый скилл ведения задач и жизни проекта: идеи/туду → планы/roadmap → раны → артефакты → прогресс/уроки. Use when user says туду, запиши, идея, backlog, потом, план, спланируй, планируем, пока план, не запускай, обсудим, roadmap, этапы, приоритеты, прогресс, урок, итоги — or

BOOST
From plugin
claude-lane-stack
12464 skills24 agents5 commands12 hooks
+1
Install
$ npx -y skills add VKirill/claude-lane-stack --skill project-life --agent claude-code

How it fires

How this skill gets triggered: by you, by Claude, or both.

  • Fires itselfAuto-invocation. Claude auto-loads it when your prompt matches the work.Auto-invocation is when the right skill fires by itself at the right moment, driven by a FLOW.md router and a hook, instead of you invoking it by name. It is the difference between a skill being installed and a skill actually getting used.Read the full definition →
  • You can call itInvoke it directly when you want it.
  • Slash command/project-life

Context preview

The summary Claude sees to decide when to auto-load this skill.

Единый скилл ведения задач и жизни проекта: идеи/туду → планы/roadmap → раны → артефакты → прогресс/уроки. Use when user says туду, запиши, идея, backlog, потом, план, спланируй, планируем, пока план, не запускай, обсудим, roadmap, этапы, приоритеты, прогресс, урок, итоги — or

SKILL.md

project-life.SKILL.md
name: project-life
description: "Единый скилл ведения задач и жизни проекта: идеи/туду → планы/roadmap → раны → артефакты → прогресс/уроки. Use when user says туду, запиши, идея, backlog, потом, план, спланируй, планируем, пока план, не запускай, обсудим, roadmap, этапы, приоритеты, прогресс, урок, итоги — or todo, plan, planning, lesson, progress. Cold-start «где мы / handoff / продолж» is resume-project, not this skill. `$ARGUMENTS` exactly info prints the card."
argument-hint: "[info]"

Project life — one skill for tasks & project memory

> **Inside a Lane Pilot chat** (`LANE_PILOT_AGENT_TYPE` set / tools `lane_pilot_*` present) the PM does not write PROGRESS, ROADMAP or plan ticks after a run: Lane Pilot's project-life helper does it (PROGRESS ≤40 lines, one CHANGELOG line, commit `chore(progress)`). The PM keeps `.agents/todos/`, `.agents/plans/` and decision drafts. «делай» = `lane_pilot_dispatch_writer` (the whole plan at once, order through `depends_on`; no waves). A lesson = `lane_pilot_lesson` (rule, audience pm, writer or both — required; `always` only for a rule true of every writer task). Project facts that cannot be derived live in the hub memory (`lane_pilot_memory_context`, `lane-memory search|context`). `run-init`, `run-finalize`, `memory-scribe`, `resume-project` CLI and `~/.agents/bin/*` below are the terminal harness.

Info

If `$ARGUMENTS` is exactly `info`, print `references/info.md` verbatim (Russian), then stop. Do not create a todo or plan unless the user already asked for that in the same message.

Everything lives in files under `.agents/` (project) or `~/.agents/` (global). **Durable files: English only.** Chat may be Russian — answer in RU, write EN to disk. Never put secrets in any of these files. **No orchestrator MCP.** No `todo_add` / task CLI for ideas.

The lifecycle (one mental model)

idea ──► todo ──► plan ──► run ──► merged work
        (backlog) (map)   (execution)   │
                                        ▼
              memory: PROGRESS · lessons→hub · decisions · session-log · artifacts

Skipping layers **upward** is fine (small fix: idea → run). Skipping the memory update after work is not.

Folder map

<project>/.agents/
  todos/            # ideas backlog        → references/todos.md
    INDEX.md
    items/<YYYY-MM-DD-slug>/{README.md, AGENT.md, meta.yaml}
  plans/            # delivery map         → references/plans.md
    ROADMAP.md
    items/<YYYY-MM-DD-slug>/{PLAN.md, meta.yaml, artifacts/}
  runs/             # execution contracts  (skill lane-contract)
    <slug>/artifacts/<task>/...
  PROGRESS.md       # now / blocked / next → references/memory.md
  agent-notes/OPEN.md
  session-log/      # hooks own it — read only
  decisions/        # decision drafts; the docs pass publishes them to docs/decisions.md
<project>/.agents/plans/       # long-form strategy (NOT this layer)
<project>/docs/                # code documentation only, kept by the docs pass

Choose project root when cwd has `.git`/`package.json`/`CLAUDE.md`; use `~/.agents/...` when user says «глобально» or cwd is home.

When to apply what

| User says (≈) | Do | Reference | |---------------|----|-----------| | «занеси в туду / запиши / потом / не теряй» | Create/update todo item + INDEX | `references/todos.md` | | chat about a todo, no «запиши» | Do not invent a todo; ask if they want it filed | — | | «покажи туду / что в бэклоге» | Read INDEX + open items, answer in RU | — | | «закрой / сделано / не надо» | status done/dropped + INDEX | `references/todos.md` | | «планируем / пока план / не запускай / обсудим / спланируй / разбей на этапы» | **Planning session** — draft in `.agents/plans/`, no run | `references/plans.md` | | «архитектор / новое приложение / новый сервис / спроектируем продукт» | Load **app-architect**. Same plan folder, living `artifacts/`. No run | skill `app-architect` | | «что дальше по проекту» (already in session) | PROGRESS → ROADMAP → todos INDEX counts | — | | «делай / реализуй / в работу / запускай ран» | Exit planning. Lane Pilot: `lane_pilot_dispatch_writer`. Terminal: spawn a run (`lane-contract`). Link it from the plan | — | | run закончился зелёным | Terminal: tick plan task → refresh ROADMAP → rewrite PROGRESS. Lane Pilot: its helper does this | `references/memory.md` | | поправили тебя / наступил на грабли | One lesson. Lane Pilot: `lane_pilot_lesson` with audience. Terminal: `lane-memory lesson "<rule>" --for pm\|writer\|both` (hub, not a file) | `references/memory.md` | | зафиксировано крупное необратимое решение | draft in .agents/decisions/<date>-<slug>.md (the docs pass publishes it to docs/decisions.md) | `references/memory.md` | | «итоги / конец сессии» | PROGRESS current, ideas filed as todos, no orphan runs | `references/memory.md` | | «что делали / почему так / покажи отчёт» | Read session-log INDEX, run artifacts, findings — don't write them | — | | «где мы / handoff / продолж» (cold start) | Skill `resume-project` (terminal: `~/.agents/bin/resume-project .`), not this file | — |

Planning session

Stay here until the user says «делай / реализуй / в работу / запускай ран».

Triggers: «планируем», «пока план», «не запускай», «обсудим», «спланируй», «разбей на этапы», «только карта», «пока не в ран».

1. Create or open `.agents/plans/items/<YYYY-MM-DD-slug>/` (`status: draft`). 2. Each decision → edit `PLAN.md` + dated **History**. Chat is not the store. 3. Screenshots / notes → optional `items/<slug>/artifacts/`; link from PLAN **Links**. 4. Ask the next scoped question. Do not invent a run or score a conveyor.

Forbidden until «делай»:

  • `run-init`, `run-controller`, `run-supervisor`, writers
  • Claude Code **Plan mode** and any write under `~/.claude/plans/`
  • owns_paths / task YAML in this `PLAN.md`

If the host is already in Claude Plan mode: tell the user to switch to **default** and keep writing under `.agents/plans/`. Do not copy the harness file as the canon (you may quote it o

Read more
Ships withclaude-lane-stack

Claude Lane Stack is a self-hosted toolkit for coordinating coding agents through CLI processes, task contracts, Git workspaces, and verification receipts (bin/run-controller:117-159, schemas/task-v2.schema.json:1-60).

Get the whole plugin

Other skills on claude-lane-stack.