Skip to content
Agentmart
Submit a tool

task

Make a small change to a shipped product — a bug fix, behavior tweak, or micro-feature — as one verified slice, syncing the affected feature spec on the way…

Skill by RubenGlez

IndexedMIT★ 1
Write a review

Chalk Labs

Need help deploying task?

Chalk Labs sets up agents like this inside real teams: connected to your tools, tested on your data, with a human in the loop where it matters.

Book a call

Install

Works with Claude Code, Claude apps, Codex, Cursor, Gemini CLI, GitHub Copilot, VS Code, OpenCode, Goose, Windsurf. Read the repository before you give any tool access to your data.

npx agentmarket install skill/rubenglez-task

Asks which of your tools to add it to (Claude Code, Codex, Cursor, VS Code, Windsurf, Gemini CLI, Claude Desktop) and shows the change first. See https://agentmart.dev/cli.

Get the weekly diff

The weekly diff: one email a week with new agent tools, notable updates and grade changes, plus security alerts for tools you follow. Double opt-in, one-click unsubscribe.

We use your email only for this newsletter. Privacy

About

Make a small change to a shipped product — a bug fix, behavior tweak, or micro-feature — as one verified slice, syncing the affected feature spec on the way out. Use for day-to-day requests like "fix X", "change Y to do Z", or a single small feature that doesn't warrant a full phase of the workflow.

Use cases: productivity

No reviews yet. Installed it? Reviews come from verified installers only. How reviews work

Permissions

What this skill can do and what it asks for, from its SKILL.md. Checked Oct 25, 2026.

Tools
No tools found
Secrets
None asked for
Install scripts
None found

Network hosts it talks to 1

  • codeload.github.com · found in install command

Files it touches

  • ~/.claude/skills/task

What's inside

Skill name
task
Description
Make a small change to a shipped product — a bug fix, behavior tweak, or micro-feature — as one verified slice, syncing the affected feature spec on the way out. Use for day-to-day requests like "fix X", "change Y to do Z", or a single small feature that doesn't warrant a full phase of the workflow.
License
MIT
Instructions
54 lines in SKILL.md · view on GitHub

Files (1)

  • SKILL.md

SKILL.md (first lines)

# Task

One small change, done end-to-end: classify, fix, verify, sync the docs. No interview, no phase ceremony. The point is that `.harness/` stays truthful as a side effect of doing the work.

## Step 1: Classify and scope

Restate the request in one sentence. Then classify:

- **Bug** — the product violates its own spec. The spec is right; the code is wrong.
- **Behavior change** — the spec said one thing, the user now wants another.
- **Micro-feature** — new behavior, small enough that a spec interview would be ceremony.

Find what's affected: read the relevant feature section in `.harness/engineering/features.md`, plus `AGENTS.md` and `.harness/product/CONTEXT.md` if they exist.

**Escalate instead of proceeding** if any of these turn out true — recommend the right skill and stop:
- It's a meaningful new feature or behavior change that warrants a product conversation → `/evolve`
- It needs an architectural decision not covered by `architecture.md` or ADRs → `/evolve` then `/dev-plan`
- It spans several features or layers in non-trivial ways → `/implement` with proper specs
- It has security, auth, data-migration, or external-contract implications → surface the question to the user first

## Step 2: Fix as one vertical slice

Implement only this change, entry point through to storage:
- Match existing conventions and use the domain vocabulary from `CONTEXT.md`
- No refactors, no shared infrastructure, no adjacent improvements
- If the change contradicts an acceptance criterion, that's expected for a behavior change — Step 4 fixes the spec, not the other way around

## Step 3: Verify

Batch size is one, so verify directly (the /implement-vs-/qa split doesn't apply here):
- Run static checks (typecheck, build, lint)
- Run the narrowest test that covers the change; for a bug, write the failing test first if the suite makes that cheap
- Exercise the change where it runs if a runtime is available (browser, simulator, CLI)

Not verified → not done. Don't hand unverified changes back.

## Step 4: Sync the docs

- **Bug**: spec was already right — no spec change. Note the fix only if the spec's implementation note is now misleading.
- **Behavior change**: patch the acceptance criterion in the affected spec to describe the new behavior (keep the Given/When/Then phrasing), with a dated one-line note ("Changed 2026-06-10: due date now resets on duplication").
- **Micro-feature**: add a minimal spec section to `.harness/engineering/features.md` — status `done`, the acceptance criteria you just verified phrased Given/When/Then, one-line implementation note. This keeps future `/qa` runs covering it.
- Touch `.harness/product/roadmap.md` only if the change alters something the roadmap states.
- If a README or other public doc describes the old behavior, update it.
- Commit the spec sync with (or right after) the code change — worktrees and future sessions only see committed `.harness/` content.

## Step 5: Report and recommend

Report: what changed (files), how it was verified, which docs were synced.

- User-visible change on a released product → "Run /ship for a patch release."
- Internal-only → done, nothing else needed.
- Found related problems along the way → list them; don't fix them in this task.

From RubenGlez/harness, MIT.

More from RubenGlez/harness

Signals

Stars
1
Forks
0
Open issues
0
Last commit (this folder)
Sep 12, 2026 (1 month ago)

From GitHub (fetched Oct 24, 2026).

Details

License
MIT
First seen
Oct 24, 2026
Last checked
Oct 24, 2026

More from RubenGlez

Done for you

Rather not set it up yourself?

Chalk Labs scopes, builds and runs agents for your team: the task audit, the eval harness, the integrations and the weekly ops report. Fixed scope, priced before we start.

See how we build AI agents →Read the case studies

Chalk Labs is a two-founder AI and growth studio. We build AI agents, CRMs and MVPs for Web3 and AI companies, and have shipped 50+ product and token launches.

launches shipped
50+
launches shipped
media placements
500+
media placements
founders on every account
2
founders on every account

Press we've placed: Forbes · CoinDesk · Decrypt · Cointelegraph · The Block

Rahil Jain, Co-founder, Chalk Labs. Ex-Goldman Sachs engineer. You talk to the person who builds it.

Get task running in your business

Tell us what you want to automate. A founder replies within one working day with a plan, not a pitch deck.

Goes to the Chalk Labs founders, not the newsletter. Privacy

Need help deploying task?

CallGet help
Cite this page

task skill. Agent Mart. https://agentmart.dev/skills/rubenglez-task (last updated 2026-10-24)