agentleFS
symonhesymonhe/homebrew-club/skills/backlog/SKILL.md

backlog

Skill1 starsChanged 3 months agoMITView on GitHub

Capture and rank a per-project backlog of work you want an agent to do unattended later. Use inside a project repo when you say "backlog this", "add to the backlog", "queue this for later", "have the runner do this", or when giving feedback about what automated work this project needs next. Also for reranking or dropping items.

39 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use issue, worktree and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • AWSSkill

    issue

    Implement a GitHub issue end-to-end — branch, code, test, docs, and commit.

    Also in 3 projects that use this

  • MicrosoftSkill

    worktree

    Create, reuse or clean up a dedicated git worktree for working on a pull request or a private branch. Use from any PR-scoped workflow that needs an isolated checkout to keep current code and build output untouched.

    Also in 3 projects that use this

  • AWSSkill

    retro

    Conduct a retrospective on a coding session.

    Also in 4 projects that use this

  • HashiCorpSkill

    changelog

    Add a `.changelog/<PR_NUMBER>.txt` entry for a GitHub Pull Request, commit, and push.

    Also in 3 projects that use this

  • BlockSkill

    plan

    Draft a TDD, PRD, or lightweight plan in the conversation without committing it.

    Also in 4 projects that use this

  • VercelSkill

    deploy

    Deploy applications to Vercel. Use when the user says "deploy", "deploy to Vercel", "push to production", "deploy my app", or "go live".

    Also in 3 projects that use this

The file

We split this skill into the rules it gives an agent

1 rule. 0 are common, 0 are distinctive, and 1 is very specific to this project.

  • 0

    Common

    Most of the 38 edited versions we read keep these.

  • 0

    Distinctive

    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:

    If a verify is not inferable and they are clearly mid-flow, mark the item needs-shaping and move on.

    line 28

“Common” is measured: kept by at least 6 in 10 of the 38 edited versions of this file we could read. “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

backlog

Capture is cheap. Execution is expensive. This skill is the cheap half.

It maintains .claude/backlog.md in whatever repo you are standing in: a ranked list of work you want done unattended, in your words, captured fast enough that you do not lose your train of thought. The burndown skill drains it later.

The backlog is unbounded. Put everything in it. The execution queue that burndown builds is capped at 3 to 5 items. That asymmetry is the whole design: capturing an idea should cost nothing, and committing to do one should cost something.

Capture mode (the default)

  1. Take the feedback as given. One item, a rapid-fire list, or a vague direction. Record each one near-verbatim in its ask: line, dated. Do not improve their wording; you will want to know what they actually said.
  2. Shape each item toward the gate below. Infer do, verify, bounds and size from the repo, and propose them. Confirm in one batch at the end, never one question per item. Interrogating someone mid-capture is how you train them to stop capturing.
  3. If a verify is not inferable and they are clearly mid-flow, mark the item needs-shaping and move on. Shape it later.
  4. An item that can never run unattended gets needs-human and one line of why. It stays visible; burndown will never take it.
  5. Rank where they say. Unstated means bottom. On a rerank or drop, edit in place and keep dropped items with their reason.

The autonomy gate

An item is only ready if all four are true. This is the same gate burndown re-checks before it runs anything, and it is the reason the whole thing is safe to leave alone.

  • No judgment call. Taste, naming, product decisions, anything where you would want a say. If a reasonable person could do it two ways and you would care which, it is not ready.
  • Reversible. The worst outcome is fixed by git restore or git revert.
  • Has a programmatic verify. A command that exits non-zero on failure. Not "check that it looks right". If you cannot name the command, you do not yet know what done means.
  • Bounded. Named paths the work stays inside.

Never queue, regardless of how it is written up: anything needing credentials or secrets, anything that sends outward (email, posts, messages, deploys, pushes), anything touching money, and anything inside a path you have marked protected.

File format

# Backlog: <repo name>
Feeder for burndown. Drained top-down; higher = sooner.

## B1: <imperative title>  [ready]
- added: 2026-08-07   ask: "<your words, near-verbatim>"
- do: <steps, with a stated default for every decision>
- verify: `<command>` (timeout 600s); pass = <observable outcome>
- bounds: <paths this may touch>
- size: S | M | L

Statuses: ready (promotable), needs-shaping (captured fast, shape later), needs-human (fails the gate by nature), promoted (moved into the queue; keep the date), dropped (with a reason). B-numbers are never reused, so a reference in a commit message stays meaningful forever.

Sizes are time caps, not estimates: S 15 min, M 45 min, L 90 min. Anything bigger is not one task.

Rules

  • Never invent backlog items. This file holds only what you were actually asked for. An agent that pads its own to-do list is an agent nobody trusts.
  • The most recent instruction wins on ordering. No silent reranking.
  • A bare ask: line with needs-shaping is a valid entry. Never let shaping questions slow down a capture burst.

Setup

Drop this file at ~/.claude/skills/backlog/SKILL.md. Commit .claude/backlog.md in each repo so the list survives a fresh clone.

Pairs with the burndown skill, which drains it.

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.