agentleFS
degdeg/beads-utils/skills/review-titles/SKILL.md

review-titles

Skill2 starsChanged 6 days agoMITView on GitHub

Review a project's bead titles against the title rules and retitle them one at a time, the user answering each in a word. Use for /review-titles, or when the user asks to review, audit, clean up or fix the titles of the open beads.

The file

We split this skill into the rules it gives an agent

6 rules. About 0 are common, 2 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

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

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

    Every title is judged against the rules in title-rules.md, beside this file.

    line 8
  2. Distinctive:

    If the set holds more than about 25 beads, say how many, and offer to narrow it by parent or priority before starting.

    line 42
  3. Very specific:

    Before showing each bead, run bd show <id> again.

    line 63
  4. Very specific:

    ok: run bd show <id> once more, then bd update <id> --title="<new title>", then show the next.

    line 78
  5. Distinctive:

    skip: leave it, and go on.

    line 82
  6. Very specific:

    Ask the user. If they are not sure either, pick the more likely reading and append a note saying so: bd update <id> --append-notes="Title …

    line 89

≈ “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

Review bead titles, one at a time

Every title is judged against the rules in title-rules.md, beside this file. Read it now. Those rules are the single source; do not restate or extend them here or in what you show the user.

The user decides each title. Your part is to check, propose and apply what they approve.

1. User- and project-specific context

Extensions are read from two places, user first, then project, and both are followed: ~/.claude/skill-extensions/review-titles.md, then .claude/skill-extensions/review-titles.md in the project. If neither exists, skip this step.

2. List the beads

The argument picks the set. Each source prints a JSON array of beads:

  • None: the live beads, open or in progress and not parked behind a deferred bead. These are the titles read every day. bd-find-live --lean names the parked beads it left out on stderr; pass that on to the user.
  • A parent id: its children, bd list --parent <id> --status=open,in_progress --limit 0 --json, run again for any child that has children of its own.
  • Several ids: those beads, bd show <id> <id> ... --json.
  • all: deferred and parked beads too, bd list --status=open,in_progress,blocked,deferred --limit 0 --json.

Pipe the source through one filter, which puts each title's length first:

jq -r '.[] | "\(.title|length)\t\(.id)\t\(.status)\t\(.title)"'

Take every length from a command, never by eye: the length is often what the user is deciding on. For a proposed title, jq -rn --arg t "<title>" '$t | length'.

If the set holds more than about 25 beads, say how many, and offer to narrow it by parent or priority before starting. Twenty-five one-word answers is already a long sitting.

3. Check every title before proposing anything

For each bead, read its description, not just its title (bd show <id>). Then check the title's claims against the code and the project's design docs. A title can read well and still be wrong about what the bead is: an "A/B test" that is an offline comparison, a "library" for a page that already exists. This is where the review earns its keep, so do it for every bead, including the ones whose titles look fine.

Then sort the beads in two groups: titles that meet the rules and are kept, and titles to change.

4. Say up front what is kept

One list, id and title, so the user can object to a keep before the one-at-a-time part starts. If they do, that bead joins the ones to change.

5. Show the rest one at a time

Before showing each bead, run bd show <id> again. If its status, assignee or title changed since the listing (another session claimed it, closed it or retitled it), say so, and ask whether to go on with it.

Show:

<id>  (<status>)
  now:      <current title>   (<length>)
  proposed: <new title>       (<length>)
  why:      <one short clause: which rule, or what the title gets wrong>

The user answers in a word:

  • ok: run bd show <id> once more, then bd update <id> --title="<new title>", then show the next.
  • change, or a question about a word: offer two or three alternatives, each with its length, and say which you recommend and why. Apply the one the user picks.
  • skip: leave it, and go on.

One bead per message. Do not batch them, and do not apply a title the user has not seen in its final form.

6. When a bead's intent is unclear

Ask the user. If they are not sure either, pick the more likely reading and append a note saying so: bd update <id> --append-notes="Title review: intent unclear, read as <reading> because <reason>." Use --append-notes, never --notes or --description; those replace the field, and the original text must stay.

7. Finish

Report the counts: kept, retitled, skipped. List any bead that changed under you during the review, and any whose notes now record an unclear intent.

Retitling writes only beads data. This skill commits nothing and pushes nothing; pushing the beads data is the user's call.

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.