agentsclimarketplace

Work batch

Skill christopherlouet/claude-base/.claude/skills/work-batch

Opinionated Claude Code foundation — Explore → TDD → Audit workflow, auto-detected stack presets (nextjs, fastapi, astro, ...), curl | bash install. MIT.

Install
npx -y skills add christopherlouet/claude-base --skill work-batch

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

  • 5 stars5 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Sequential execution of user stories from a PRD file. Autonomous mode that implements and commits each story one by one. Trigger when the user wants to process a backlog, execute multiple stories, or launch an autonomous mode.

SKILL.md

2.6 KB, as published. Nobody here has run it

Batch Execution Mode

Autonomous and sequential execution of user stories from a PRD file.

PRD file format

JSON (prd.json)

{
  "project": "project-name",
  "stories": [
    {
      "id": "US-001",
      "title": "Story title",
      "description": "Detailed description",
      "priority": "P1",
      "acceptance_criteria": [
        "Given X, When Y, Then Z"
      ],
      "files": ["src/module.ts", "src/module.test.ts"]
    }
  ]
}

Markdown (prd.md)

## US-001: Story title
**Priority**: P1
**Description**: Detailed description
**Acceptance criteria**:
- Given X, When Y, Then Z
**Files**: src/module.ts, src/module.test.ts

Workflow per story

For each story in priority order (P1 → P2 → P3):

1. LOAD - Load the story

  • Read the story from the PRD file
  • Display the title and description
  • Check prerequisites (files, dependencies)

2. IMPLEMENT - TDD cycle

  • Write the tests first (RED)
  • Implement the minimal code (GREEN)
  • Refactor if necessary (REFACTOR)
  • Verify the acceptance criteria

3. COMMIT - Save

  • Run the tests: npm test / pytest / go test
  • If tests OK: commit with feat(scope): US-XXX description
  • If tests KO: STOP and report the blocker

4. REPORT - Update the state

  • Mark the story as done in .claude/output/batch/progress.json
  • Log the time and modified files
  • Move on to the next story

Progress file

Automatic save in .claude/output/batch/progress.json:

{
  "started_at": "2026-03-23T10:00:00Z",
  "stories": {
    "US-001": { "status": "done", "commit": "abc1234", "files": ["..."] },
    "US-002": { "status": "in_progress" },
    "US-003": { "status": "pending" }
  }
}

Resume after interruption

If progress.json exists, resume at the last in_progress or pending story.

Guardrails

  • Maximum 10 stories per batch (beyond that, split)
  • STOP if 2 consecutive stories fail
  • Each story must pass the tests before continuing
  • Commit after each story (no giant commits)

IMPORTANT: Each story is an atomic commit. Do not accumulate changes.

NEVER continue if the tests fail on a story.

YOU MUST save the progress in .claude/output/batch/progress.json.

Keep looking

Skills are one crate of 328,083. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.