agentleFS
rome-osrome-os/rome/.claude/skills/todos-to-github-project/SKILL.md

todos-to-github-project

Skill904 starsChanged 4 days agoMITView on GitHub

Turn the remaining todos of the current conversation into a GitHub project of task-spec issues wired with native blocked-by dependencies.

The file

We split this skill into the rules it gives an agent

8 rules. About 0 are common, 2 are distinctive, and 6 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.

  • 6

    Very specific

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

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

    Search the tracker for each todo with gh issue list --search "<keywords>".

    line 16
  2. Distinctive:

    Write one sentence per todo stating the outcome its issue must deliver.

    line 17
  3. Distinctive:

    Check the edge graph for cycles.

    line 25
  4. Very specific:

    If gh label list does not show task or ready-for-agent, create the missing label with gh label create.

    line 34
  5. Very specific:

    Create the project: gh project create --owner <owner> --title "<title>".

    line 37
  6. Very specific:

    Create the issues in dependency order, blockers first, so every Blocked by line names a real issue number: gh issue create --title …

    line 39
  7. Very specific:

    Add each in-scope issue to the project: gh project item-add <project-number> --owner <owner> --url <issue-url>.

    line 40
  8. Very specific:

    Read back each issue's edges with gh api repos/<owner>/<repo>/issues/<number>/dependencies/blocked_by.

    line 51

≈ “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

Todos to GitHub project

Goal end-state: every remaining todo of the conversation is one task-spec issue, the issues sit in one GitHub project, every dependency is a blocked-by edge on GitHub, and each in-scope issue carries ready-for-agent.

The issue format — title, body sections, the ready-for-agent bar — is defined in docs/authoring/github-issues.md and docs/authoring/github-issues-task-spec.md. Prose in titles and bodies follows docs/authoring/WRITING.md. This skill covers what to file, in which order, and with which gh calls.

1. Harvest

  1. Collect every remaining todo from the conversation: open task-list items, deferred follow-ups, and work the user parked for later.
  2. Search the tracker for each todo with gh issue list --search "<keywords>". If an existing issue already tracks it, drop the todo from the set.
  3. Write one sentence per todo stating the outcome its issue must deliver. If the sentence only makes sense with the conversation open, ask the user what the todo means.

2. Draft

  1. Name the project after the outcome the set delivers, not after the conversation.
  2. Split the todos into in-scope and out-of-scope. An out-of-scope todo files under its issue type with the type label, stays out of the project, and never carries ready-for-agent.
  3. Draft each in-scope issue as a task spec in the github-issues-task-spec.md format, title included.
  4. Wire the dependencies. Issue X is blocked by issue Y when X cannot start before Y merges. Record each edge as a Blocked by line in X's body.
  5. Check the edge graph for cycles. A cycle is a decomposition error — re-split the todos until the graph is acyclic.
  6. Apply the label bar from github-issues-task-spec.md: every in-scope issue gets task, and an issue implementable from its body alone also gets ready-for-agent. An issue still waiting on a user decision files without ready-for-agent.

3. Approve

The draft stays out of chat. Ask for the go-ahead in one line — the project title and the issue count — and write nothing to GitHub before the user gives it. Present issue titles or bodies only when the user asks to see them.

4. Create

If gh label list does not show task or ready-for-agent, create the missing label with gh label create.

  1. Resolve the repo and owner: gh repo view --json nameWithOwner.
  2. Create the project: gh project create --owner <owner> --title "<title>".
  3. Link the project to the repo so it shows up under the repo's Projects tab: gh project link <project-number> --owner <owner> --repo <owner>/<repo>.
  4. Create the issues in dependency order, blockers first, so every Blocked by line names a real issue number: gh issue create --title "<title>" --body "<body>" --label task --label ready-for-agent. Drop ready-for-agent where step 2.6 withheld it.
  5. Add each in-scope issue to the project: gh project item-add <project-number> --owner <owner> --url <issue-url>.
  6. Mirror every edge into GitHub's native relation. The issue_id is the blocking issue's numeric REST id (gh api repos/<owner>/<repo>/issues/<number> --jq .id), not its issue number:

    gh api -X POST repos/<owner>/<repo>/issues/<blocked-number>/dependencies/blocked_by -F issue_id=<blocking-id>

5. Verify

  1. Confirm the link with gh repo view --json projectsV2. The new project must appear in the list.
  2. List the project items with gh project item-list <project-number> --owner <owner>. The count must match the in-scope set.
  3. Read back each issue's edges with gh api repos/<owner>/<repo>/issues/<number>/dependencies/blocked_by. Every drafted edge must appear.
  4. Report the project URL and each issue number with its blockers.

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.