Skip to content
Skills Library
AI & agents · Building & coding

perform-task

/perform-task

Implement, self-review, test, and commit exactly one existing ai/tasks task, by slug or full date/slug id. Use when the user invokes /perform-task <name>, or the continue scheduler delegates one selected task.

Repository
harness ↗
Licence
MIT
Last repository change
12.11.2026

Assessment and security

100/ 100Class: good
Format25 / 25
Frontmatter with name and description, valid name, description up to 1,024 characters, name matching the folder.
Discoverability25 / 25
Description explains when to use the skill, has a reasonable length and uses third person.
Content35 / 35
At least 150 words and at most 500 lines, headings, step lists, examples or code.
Hygiene15 / 15
No placeholders such as TODO, no Windows paths, with a licence.
Riskno signals detected in the scan
Files in this package
1
Scripts
0
HTML comments
0
URLs in SKILL.md
0
Deprecation

No signals detected. Repository changed on 12.11.2026, without references to retired models or APIs.

Licence

MIT, open: use and modify while retaining author attribution.

What it does and when to use it

Implement, self-review, test, and commit exactly one existing ai/tasks task, by slug or full date/slug id. Use when the user invokes /perform-task <name>, or the continue scheduler delegates one selected task.

Original author description

Implement, self-review, test, and commit exactly one existing ai/tasks task, by slug or full date/slug id. Use when the user invokes /perform-task <name>, or the continue scheduler delegates one selected task.

How it works

  1. Resolve
  2. Implement
  3. Self-review
  4. Test
  5. Publish
  6. Report
Original SKILL.md headings · 76 lines, 486 words
SKILL.md preview
---
name: perform-task
description: Implement, self-review, test, and commit exactly one existing ai/tasks task, by slug or full date/slug id. Use when the user invokes /perform-task <name>, or the continue scheduler delegates one selected task.
---

# Perform One Task

Own exactly one task through to a committed result or an honestly recorded blocker. Do not
switch to a different task, do not process the inbox, and do not start a second task after
this one finishes — that's `continue`'s job.

## Resolve

Find the task under `ai/tasks/` by slug or full `date/slug` id. If the name matches more
than one task, list the matches and ask which one instead of guessing. Read that task's
`task.md` and `state.yaml`.

- `status: approved` → report the existing result and stop; don't redo it.
- `status: blocked` → read why before retrying. If the blocker is resolved, continue from
  where it left off rather than restarting from scratch.
- `status: todo` or `in-progress` → proceed.

If `depends_on` names a task that isn't `approved` yet, stop and report which one is
missing.

Set `status: in-progress` in `state.yaml` before writing any application code (fold this
into the eventual work commit if that's simpler — just don't leave the queue showing
`todo` while you're actively working on it).

## Implement

Read `AGENTS.md` and `.agents/shared/checks.md` completely before writing code. Make the
smallest change that satisfies every line under `task.md`'s `## Acceptance`. Stay inside
this task's own scope: a genuinely separate improvement you notice along the way goes into
`ai/inbox.md` as a new note for a future task — it does not ride along in this diff.

## Self-review

Read the complete diff yourself before running any checks. Walk it against `task.md`'s
Acceptance criteria one by one. Look for obvious correctness issues and confirm the diff
doesn't touch files the task had no reason to touch.

## Test

Run every check in `.agents/shared/checks.md`. A check that fails only on something listed
under `AGENTS.md`'s "Known baseline issues" doesn't block this task; anything your own
change caused does. If a check fails because of this task's change, fix it and re-run —
never report a task done with a failing check that your diff caused.

Append a `## Result` section to `task.md`:

```markdown
## Result

- Commands run: <exact commands>
- Lint: <pass | pre-existing baseline errors only, listed>
- Build: <pass | dev-mode fallback used, why>
- Smoke tests: <pass | which spec(s) added/extended>
- Screenshot: <path under .agents-runs/<task-id>/, for a UI change — or "n/a">
```

## Publish

```bash
git add <exact files this task touched, plus ai/tasks/<id>/task.md and state.yaml>
git commit -m "<Imperative summary>

Task: <task-id>"
```

Set `status: approved` in `state.yaml` in the same commit — or `status: blocked` with the
exact reason written into `task.md`, leaving the branch otherwise clean and the task
resumable. Never commit a file this task didn't touch, and never commit anything under
`.agents-runs/` (it's gitignored on purpose — scratch evidence, not project history).

## Report

Task id, final status, files touched, checks run and their results, and anything you
noticed but deliberately left out of scope (and confirmation it's now in `ai/inbox.md`).

Versions

The database has no other skills named perform-task.

Database snapshot: 13.11.2026 · package version cee748706e12 (SHA-256 digest of the entire package)

Downloads in the last 30 days: 0 · ranking

Report a problem with this skill: harmful content, incorrect descriptions or an author’s removal request.

Similar skills

  • 100do-work
    Execute the approved plan in work/<slug>/README.md (status ready), write outputs into work/<slug>/data/, record what happened in the Results section, then hand off to review-work. Use when there is a ready work item.
  • 100implement-tasks
    Use to implement a feature from its tasks.json with test-driven development — writes a failing test first, makes it pass, refactors, gates, and commits per task. Triggers on "/sdlc:implement-tasks {slug}", "implement {slug}", "build {slug}", "TDD {slug}", "code up the tasks for…
  • 100break-tasks
    Use to break a designed feature into atomic, ≤1-day tasks with a dependency graph, a per-task Definition of Done, and a machine-readable tasks.json that the implement-tasks engine consumes. Triggers on "task breakdown for {slug}", "break down tasks for {slug}", "tasks for…
  • 100loading-addressable-markdown
    Use when a mission, slice, dashboard, or task references Markdown as path#h2-slug or path#h2-slug/h3-slug outside the OpenRig context-pack library.
  • 100implement
    Use to implement a feature from its tasks.json with test-driven development — writes a failing test first, makes it pass, refactors, gates, and commits per task. Triggers on "implement {slug}", "build {slug}", "TDD {slug}", "code up the tasks for {slug}", "/sdd:implement…
  • 100tasks
    Use to break a designed feature into atomic, ≤1-day tasks with a dependency graph, a per-task Definition of Done, and a machine-readable tasks.json that the implement engine consumes. Triggers on "task breakdown for {slug}", "break down tasks for {slug}", "tasks for {slug}",…

Feedback and reports

Your message goes to Bartosz Sokoliński. An email address is needed only for a reply. See the privacy policy.