agentleFS
BSSE23047BSSE23047/nauraiz-dev/skills/kickoff-prompt/SKILL.md

kickoff-prompt

Skill2 starsChanged 15 days agoMITView on GitHub

Write a kickoff prompt for an issue or task, for a fresh session here or elsewhere: guard block, branch, files, decisions, gate, stops, model, budget. Use when asked for a kickoff prompt. Not for doing the task.

The file

We split this skill into the rules it gives an agent

1 rule. About 0 are common, 1 is distinctive, and 0 are 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.

  • 0

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Write one prompt that a fresh session can run without this conversation.

    line 9

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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

kickoff-prompt

Write one prompt that a fresh session can run without this conversation. Do not start the task. Get the shape with node "${CLAUDE_PLUGIN_ROOT}/scripts/show.mjs" templates/kickoff-prompt.md (never Read plugin files directly), keep its guard block first, and fill every {{…}}; none may remain.

Inputs

  1. The issue: gh issue view <n> --json title,body,assignees,labels,comments. A sentence instead of an issue: that sentence is the spec. gh fails: say so and ask for the text.
  2. Config: .claude/nauraiz-dev.json, with .claude/nauraiz-dev.local.json over it: branch.base, branch.pattern, gate, models, protectedPaths, mockupRoot.
  3. git rev-parse --show-toplevel and git branch --show-current.

Filling the template

  • {{absPath}}: the absolute directory the run must work in, in the OS's own form: the repo root, or the worktree the owner named.
  • {{branch}}: the branch the issue names, else branch.pattern with the issue number and a short slug. {{base}}: branch.base.
  • {{version}}, {{enforcement}}: from this session's start line nauraiz-dev v… · enforcement=…. Not in your context: ask the owner.
  • {{title}}, {{issueRef}}: the issue title and #<n> (or "no issue").
  • Files to read: paths the issue names, plus the files that own the behaviour (find them with Grep; list paths, not contents). Files not to read: generated folders, lockfiles, protectedPaths, and what the issue puts out of scope.
  • Mockups and specs: the files the issue names, under mockupRoot when it is set. Never invent a path; none known: "None."
  • Decisions already made: only what the issue, its comments or the owner in this session decided, each with its source. Open questions go under "Decisions to confirm", each with a recommendation, for the run's grill.
  • Gate: one line per gate entry (name: command); none configured: say so.
  • Model: size the task small / normal / large as start-task does and name models.<size> and the size.
  • Budget: the cap the owner gave (USD or time). None given: recommend one by size (small 1 USD, normal 4 USD, large 10 USD) and mark it "(recommended, owner to confirm)".
  • {{extraStops}}: stop conditions the issue adds, one - line each, or delete the line.
  • The owner authorised a PR in this session: replace the first "Never" line with what they allowed. Otherwise keep it.

Output

Print the prompt in one fenced block, ready to paste. Under it, print one line naming the model and the budget, and one line: "Before pasting: check out <branch> in <absPath>" (the guard stops the run on any other branch or directory). Write no file unless asked.

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.