agentleFS
paslavskyipaslavskyi/llm-wiki/.claude/skills/kb-plan/SKILL.md

kb-plan

Skill1 starsChanged 32 days agoMITView on GitHub

Use when the owner talks about the operational development plan in dialogue — "зараз робимо X", "потім Y", "відклади Z у backlog", "піднімай W з backlog", "X зроблено", "покажи план/чергу". Translates natural language into deterministic tools/plan.mjs commands; never edits the plan YAML by hand.

The file

We split this skill into the rules it gives an agent

2 rules. About 0 are common, 1 is distinctive, and 1 is very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 1

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    NEVER edit the plan note with Write/Edit or kb:write — only tools/plan.mjs (it heals, validates disjointness, reindexes, and commits …

    line 38
  2. Distinctive:

    After every operation: show the resulting lists (the CLI prints them) so the owner immediately verifies the plan matches what they said.

    line 48

≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

kb-plan — dialogue-driven operational planning

Goal: turn planning statements into deterministic plan operations. The mechanics live in tools/plan.mjs; this skill ONLY translates language → commands.

Model

  • PLAN-001 (singleton, type plan): active (в розробці), queue (TODO, ORDER = execution sequence), backlog (parked).
  • done is never stored — it is derived from the plan note's git history. That is why every operation makes its own commit with a canonical message.
  • Goals (GOAL-) are independent of the plan; a plan item may be any live note id (MIL/FEAT/FR/RISK/Q/GOAL/...).
  • Dependencies live on the items, not the plan: relations: { depends_on: [ids] } (via kb-capture; cycles fail validate). A dependency is met when its note is accepted, deprecated, or done per plan history. show marks blocked queue items (⛔ чекає: …) and prints НАСТУПНЕ ГОТОВЕ; take with no id starts that item, not the queue head; take X on a blocked item warns — tell the owner what it waits on. «X залежить від Y» → add depends_on: [Y] to X (kb-capture), not a plan verb.

Verb mapping (run via npm run plan -- <verb> ...)

Owner saysCommand
«зараз робимо X, потім Y, Z»set --active X --queue Y,Z (confirm what happens to items currently in lists but not mentioned — done/drop/backlog?)
«беремо X в роботу» / «беремо наступну»take X / take (= next READY, skips blocked)
«X зроблено / готово»done X
«X більше не актуально, прибери»drop X (NOT done — history distinguishes by message)
«відклади X у backlog / колись повернемось»defer X
«піднімай X з backlog»promote X [--pos N] (ask position if unclear; default = end of queue)
«X перед Y / підніми X вище»move X --pos N
«покажи план / чергу / що в роботі»show

Rules

  • NEVER edit the plan note with Write/Edit or kb:write — only tools/plan.mjs (it heals, validates disjointness, reindexes, and commits atomically; the commit message is load-bearing for done-history).
  • Resolve titles to ids first (kb-recall / indexes). If the owner names work that has no note yet, capture it first (kb-capture), then plan it.
  • On «X зроблено» where X is a goal or has a scope: check scope completion (npm run impact -- X, note statuses) and surface unfinished tails instead of silently closing.
  • Ambiguity (position in queue, what to do with unmentioned items in set) → ask, don't guess.
  • After every operation: show the resulting lists (the CLI prints them) so the owner immediately verifies the plan matches what they said.

Output (per mode)

  • debug: command(s) run + resulting lists.
  • autonomous: short confirmation + resulting queue.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.