agentleFS
darinhdarinh/copilot-tools/skills/backlog/SKILL.md

backlog

Skill1 starsChanged 4 months agoMITView on GitHub

Filing, approving, closing, and writing evidence for items in the tracked backlog/ directory. Load this skill before creating a backlog item, closing one, or deciding whether work belongs in the backlog or a spec.

36 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, 1 is distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 35 edited versions we read keep these.

  • 1

    Distinctive

    This author's own choices. Worth a look.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    A plausible mechanism is not a diagnosis: If you can describe how a defect works but have not demonstrated it, say which is which in the …

    line 105

“Common” is measured: kept by at least 6 in 10 of the 35 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

Open work belongs in the repository. Handoffs and session logs are handover mechanisms, not records — closed work is answerable from git log, open work is answerable from backlog/ and nowhere else.

One file per item

backlog/0007-short-kebab-slug.md — four-digit id, kebab slug.

This is the design constraint, not a style choice. Parallel agents work in separate worktrees off the same main; a single BACKLOG.md puts every add and every close on the same lines of one file, so every concurrent pair conflicts. One file per item makes an add conflict-free by construction and a close a small diff.

---
id: 7
title: One line, imperative
status: proposed
opened: 2026-08-04
closed:
commit:
spec: specs/003-a-feature/spec.md
requirement:
---

## Evidence
## Why it matters
## Notes

Front matter takes no inline comments — a value is the rest of its line. YAML would read title: Fix issue #42 as Fix issue, silently discarding the number, and a status: open # or closed copied from an example would be a status nobody can parse.

The approval gate

The vocabulary is proposed → open → closed | rejected, and the first arrow is a human's.

  • proposed — filed, not yet approved. An agent may file one of these unprompted; it may not work one. This status exists so the gate can be expressed: open alone conflates "somebody wrote this down" with "somebody decided it should be built", and a gate cannot be enforced against a distinction the data cannot make.
  • open — approved, outstanding. The one status an agent may pick up freely.
  • closed — shipped. The only status that names a commit.
  • rejected — considered and declined. A decision, recorded as one, which is why it is a status and not a deleted file: a deleted item is indistinguishable from an item nobody ever filed, so the next agent to notice the same defect files it again and the decision is paid for twice.
operator backlog new      # always files as proposed
operator backlog ready    # the items you are actually allowed to work
operator backlog approve  # the product owner's act: proposed -> open
operator backlog close    # closed with the commit, or rejected
operator backlog list | show | check | scrum | html

Use ready, not list. list shows everything including proposed; ready is the one that applies the gate.

Fields

  • spec — names a spec, or the literal none. Writing none out makes "this changes no specification" a decision somebody recorded rather than a silence that could equally mean nobody looked.
  • requirement — optional. When set, the text must occur in the named spec, so renaming or deleting the requirement turns the suite red and somebody has to revisit the item. An item pointing at a spec section that no longer exists is worse than one pointing nowhere, because it reads as though it had been checked.
  • blocks — the recorded exception. An unapproved item that blocks approved work may be worked; nothing else unapproved may be.

Closing

Set status, closed and commit in the same commit that does the work, and update the linked spec in that commit too. A close landing separately is a window in which the backlog is wrong, and commit cannot name a SHA that does not exist yet — so commit the work and amend, or close in the merge commit.

rejected takes a closed date and no commit, because nothing shipped. Demanding a SHA there forces whoever rejects an item to invent one, and an invented SHA is worse than a blank field because it looks like evidence.

Evidence

Evidence is what was measured — a reproduction, a command and its output, a mutation that ran green. An item with no evidence is a rumour, and a backlog of rumours costs every reader the time it takes to find that out.

Never seed an item you have not verified yourself. Carried-forward lists go stale: two of the four items this convention was first written for turned out to be already fixed, and recording them would have created work that did not exist.

A plausible mechanism is not a diagnosis. If you can describe how a defect works but have not demonstrated it, say which is which in the item. Handing over an undemonstrated mechanism costs the next reader the time to disprove it, and they will disprove it after believing it.

Backlog or spec?

The backlog is the queue; specs/ is the contract. An item is the right place for work that is not yet specified. When it grows into a feature, run /speckit-specify and point the item's spec at what that produced.

Enforcement

operator backlog check runs the rules; the suite runs it too. Ids match filenames and are unique, status is in the vocabulary, evidence is non-empty, a closed item's SHA resolves in this repository, and backlog/ itself is non-empty. Without that last check, deleting backlog/ turns every other rule into a loop over an empty list and the suite reports clean at the moment it stopped existing.

Prove each rule can fail. Violate it in a temp copy and watch the suite go red. A guard that cannot fire reads exactly like coverage.

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.