agentleFS
tkadauketkadauke/syrus/.claude/skills/implement/SKILL.md

implement

Skill9 starsChanged 8 days agoMITView on GitHub

Use when a task description asks for a feature or bug fix to be built in this codebase — write, verify, and commit the actual code changes, not a plan or a partial attempt.

  • Deletes or force-pushes

192 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use setup-ts-deep-modules, ask-matt 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.
  • liufeng1976Skill

    setup-ts-deep-modules

    Wire dependency-cruiser into a TypeScript repo so each package is a deep module — implementation hidden in subfolders, reachable only through its entry-point files. User-invoked.

    Also in 43 projects that use this

  • GoogleSkill

    ask-matt

    Ask which skill or flow fits your situation. A router over the skills in this repo.

    Also in 91 projects that use this

  • RoentekSkill

    loop-me

    Grill me about specs for the workflows I want to build, within this workspace.

    Also in 50 projects that use this

  • AWSSkill

    to-questionnaire

    Turn a decision you can't fully answer into a questionnaire for someone else to fill in.

    Also in 88 projects that use this

  • ryotaroh180105Skill

    claude-handoff

    Hand the current conversation off to a fresh background agent that picks up the work immediately.

    Also in 42 projects that use this

  • GoogleSkill

    to-spec

    Turn the current conversation into a spec and publish it to the project issue tracker — no interview, just synthesis of what you've already discussed.

    Also in 168 projects that use this

  • +3 more from google
  • Other versions192 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 192

    1. mattpocock/skills★ 325ksame name
    2. QuivrHQ/quivr★ 40ksame name
    3. The-Vibe-Company/quivr★ 40ksame name
    4. foambubble/foam★ 17ksame name
    5. formbricks/formbricks★ 13ksame name
    6. ArcReel/ArcReel★ 6.6ksame name
    7. vinvcn/mattpocock-skills-zh-CN★ 5.8ksame name
    8. fallow-rs/fallow★ 5.4ksame name
    9. withcoral/coral★ 5ksame name
    10. bestofjs/bestofjs★ 3.1ksame name

    182 more not listed

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

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

  • 0

    Common

    Most of the 175 edited versions we read keep these.

  • 4

    Distinctive

    This author's own choices. Worth a look.

  • 2

    Very specific

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

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

    Before you start, Syrus may run setup commands from .syrus.yml at the repo root.

    line 15
  2. Distinctive:

    If a tool gives you trouble (a setup task fails because dependencies aren't installed, a linter blows up, etc.), surface that in the run's …

    line 62
  3. Very specific:

    NOT trigger a later turn or out-of-band notification inside this Run, and ScheduleWakeup is for chat sessions, not for continuing a …

    line 82
  4. Distinctive:

    Run diagnostic and verification commands in the foreground with an adequate timeout when you need their result.

    line 86
  5. Distinctive:

    Run focused local validation only: the exact failing check, the smallest relevant test target, or a narrow command tied to the files you …

    line 92
  6. Distinctive:

    Before stopping, diff your changes against the task description.

    line 118

“Common” is measured: kept by at least 6 in 10 of the 175 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

$ARGUMENTS


Syrus context — Syrus is the automation harness running this agent inside a cloned repository. It turns GitHub issues, PR feedback, scheduled tasks, retries, and rebases into agent runs, then captures your commits and opens or updates the PR.

Before you start, Syrus may run setup commands from .syrus.yml at the repo root. Supported shape:

prepare: - bundle install - npm ci

prepare: [] or prepare: false opts out. If .syrus.yml is absent, Syrus auto-detects one setup command from common files such as Gemfile, yarn.lock, pnpm-lock.yaml, package-lock.json, or package.json. Do not edit .syrus.yml unless the task asks you to fix setup itself.

In project-aware monorepos, the environment snapshot may also list Target prepare options compiled from root or nested .syrus.yml files. Read that line before assuming the root prepare covered every project dependency. If code exploration shows you need a project-specific environment, call the audited workflow MCP tool run_target_prepare(label:, reason:) with one of the listed labels, such as //apps/web:prepare. Do not invent project labels, and do not expect Syrus to infer a project just because a package file exists; projects and targets come from explicit .syrus.yml declarations or build-system providers.

Live Syrus state — if you need to make a claim about the current Job, Workflow, Run, queue, PR, or related chat state, call the read-only read_live_state MCP tool first. Prompt text can be stale by the time you act; the tool is the approved current-state source. Do not use it to mutate jobs or queues.


Git pipeline contract — Syrus runs your work through:

commit_agent_changes → git diff origin/<default_branch>...HEAD → push → open PR

For that pipeline to work, your branch's HEAD must share history with the repo's default branch. Don't break that invariant. In particular, NEVER run any of these mid-run:

  • git checkout --orphan ... / git switch --orphan ...
  • git reset --hard <unrelated commit>
  • git rm -r . followed by re-adding everything
  • rm -rf .git && git init
  • git update-ref on HEAD or refs/heads/*
  • git commit-tree produced by yourself, then attached to HEAD

If a tool gives you trouble (a setup task fails because dependencies aren't installed, a linter blows up, etc.), surface that in the run's requested reporting channel instead of working around it via destructive git ops. If you have verified a concrete blocker means the work cannot proceed in this workflow environment (for example a required binary is missing from the worker image, or the task asks for a capability this workflow step does not have), call report_cannot_proceed(reason:, details:) and then stop work. Syrus will mark the Run, Step, and Workflow blocked instead of failed, record the reason on the Job, and surface it for operator attention.

Sane git ops are fine — git status, git log, git diff, git add, git commit, git restore, git stash (if you pop it back). git rebase and git merge against <default_branch> are also fine — they preserve history.


Shell command execution contract — this is one agentic Step Run, not an ongoing chat session. A backgrounded shell command finishing does NOT trigger a later turn or out-of-band notification inside this Run, and ScheduleWakeup is for chat sessions, not for continuing a workflow Step Run.

Run diagnostic and verification commands in the foreground with an adequate timeout when you need their result. If you deliberately background a command, you must actively poll or monitor its output in this same turn and finish interpreting it before your turn ends. Do not end the turn saying you will wait to be notified later.

Run focused local validation only: the exact failing check, the smallest relevant test target, or a narrow command tied to the files you changed. Do not run broad/full-suite validation unless explicitly instructed. Syrus will run the repository's configured graders in parallel after your step; use those for broad feedback. If no focused validation path is available, explain that and finish after code review plus targeted checks.


Corner-cutting contract — this is a one-shot, unattended run with no one to catch a shortcut before it ships. Rationalizations that show up under pressure, and why they don't hold here:

  • "The happy path works, ship it." → Run the relevant tests or build before claiming something works. An unverified claim is not a working fix.
  • "This is new/changed behavior, but it doesn't really need a test." → New or changed behavior gets a test in the same commit. No test means the behavior isn't actually defined yet.
  • "This linter config / dev dependency / tool is broken, I'll just work around it." → Same rule as the destructive-git-ops guidance above, generalized: report a broken tool instead of silently routing around it. Working around it hides a problem the operator needs to know about.
  • "I've made some progress, that's close enough to stop." → Before stopping, diff your changes against the task description. Stopping early and reporting done without checking the diff actually matches the task is not done.

Phased execution note: you're running the implement step. Make the code changes; commit them locally; that's it. DO NOT call submit_summary here. A separate, short follow-up step will ask you to summarize the work for the PR — your full context will be available to it via session resume, so you don't need to summarize ahead of time. If you finish early, just stop your tool calls and let the run end; we'll prompt you for the summary on the next step.

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.