agentleFS
jerssonjersson/jenie-ai-tools/skills/implement-task/SKILL.md

implement-task

Skill2 starsChanged 55 days agoMITView on GitHub

Implement a validated task end-to-end. Checks that analyze-task ran within the last hour, then branches, writes tests (TDD), implements, commits, lints, and commits again.

16 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use handoff and code-review.

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.
  • MicrosoftSkill

    handoff

    Compact the current conversation into a handoff document for another agent to pick up.

    Also in 3 projects that use this

  • GoogleSkill

    code-review

    Adversarial systems, concurrency, memory-model, and architecture code review protocol. Use this skill whenever asked to review code, PRs, diffs, or branches in Filament.

    Also in 3 projects that use this

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. 0 are common, 6 are distinctive, and 13 are very specific to this project.

  • 0

    Common

    Most of the 14 edited versions we read keep these.

  • 6

    Distinctive

    This author's own choices. Worth a look.

  • 13

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    When this skill is loaded, you help the developer implement a validated task (user story or bug) from start to finish — branching, TDD, …

    line 8
  2. Very specific:

    Read ../globals/INDEX.md and follow its instructions to load all globals.

    line 16
  3. Very specific:

    Check whether a .git directory exists at the project root (<project_root>/.git — use ls or test -d).

    line 20
  4. Very specific:

    Check whether <project_root>/.jenie/technical-context.md exists (use ls or test -f).

    line 24
  5. Distinctive:

    If it exists: Read the file and cache its contents as the session's technical reference.

    line 26
  6. Distinctive:

    Stop: Do not proceed with implementation.

    line 30
  7. Very specific:

    Ask the developer which task they want to implement — a file path or a reference to a task from docs/.

    line 36
  8. Very specific:

    Read <project_root>/.jenie/.validation-registry.json.

    line 42
  9. Very specific:

    If the entry's status is not "consistent-doable":

    line 47
  10. Distinctive:

    Ask if they want to refine the task and re-validate, or stop.

    line 49
  11. Very specific:

    If the entry's status is "consistent-doable" and the timestamp is within the last hour — proceed.

    line 50
  12. Very specific:

    Read ./guidelines/git.md and follow its branch naming and working tree conventions.

    line 54
  13. Very specific:

    Read ./guidelines/tdd.md and follow its TDD process: extract criteria, discover test infrastructure, write tests (red phase), then …

    line 60
  14. Very specific:

    Use search_graph(label="Function", name_pattern=".*test.*") to find existing test patterns

    line 63
  15. Very specific:

    Use get_code_snippet() to read relevant test examples

    line 64
  16. Very specific:

    Read ./guidelines/quality-check.md and run all checks — linting, type-checking, security, and manual code review.

    line 72
  17. Distinctive:

    Run the tests again — confirm all are still green.

    line 76
  18. Distinctive:

    If any regression was introduced, fix before proceeding.

    line 77
  19. Very specific:

    Follow the quality commit conventions from the cached git guidelines (./guidelines/git.md).

    line 81

“Common” is measured: kept by at least 6 in 10 of the 14 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

Implement Task Skill

When this skill is loaded, you help the developer implement a validated task (user story or bug) from start to finish — branching, TDD, implementation, linting, and committing.

Command syntax: jenie commands are written here as /jenie <name>. In OpenCode use them as-is (e.g., /jenie prepare-for-wishes); in Claude Code they are namespaced with a colon (e.g., /jenie:prepare-for-wishes); in Copilot use the skill name directly (e.g., /prepare-for-wishes). Always show the developer the syntax that matches the current environment.

Pre-condition

1. Load Global Guidelines

Read ../globals/INDEX.md and follow its instructions to load all globals. Cache them for the session. This must complete before any workflow step.

2. Verify repository exists

Check whether a .git directory exists at the project root (<project_root>/.git — use ls or test -d). If it does not — there is no repository and no code to modify — inform the developer and stop. Do not proceed.

3. Load technical context

Check whether <project_root>/.jenie/technical-context.md exists (use ls or test -f).

  • If it exists: Read the file and cache its contents as the session's technical reference. The information within (scripts, naming conventions, project structure, workflow) should guide implementation decisions throughout all subsequent steps.
  • If it does not exist: This is a blocker. Tell the developer:

    ⚠️ Technical context not found. The file .jenie/technical-context.md is required for implementation (it provides project structure, scripts, naming conventions, and workflow rules). Please run /jenie prepare-for-wishes first to generate it.

    Stop. Do not proceed with implementation.

Workflow

1. Determine the task to implement

Ask the developer which task they want to implement — a file path or a reference to a task from docs/. If they're unsure, suggest using /jenie list-tasks first to browse and stop.

Once you have the task file path, read the full task file so you know its acceptance criteria and details. If the developer does not provide a task after the prompt, inform them that implementation cannot proceed without a task and stop.

2. Check validation recency

  1. Read <project_root>/.jenie/.validation-registry.json. If the file does not exist or contains invalid JSON, no validation has been recorded.
  2. Find an entry whose path matches the task file path (relative to project root).
  3. Determine the current UTC time. If no entry exists or the entry's timestamp is older than 1 hour from the current time:
    • Tell the developer the task needs re-validation (it was never validated or the validation expired).
    • Redirect to /jenie analyze-task and stop.
  4. If the entry's status is not "consistent-doable":
    • Inform the developer the task did not pass validation (status: <status>).
    • Ask if they want to refine the task and re-validate, or stop. Do not proceed with implementation.
  5. If the entry's status is "consistent-doable" and the timestamp is within the last hour — proceed.

3. Create a feature branch

Read ./guidelines/git.md and follow its branch naming and working tree conventions. Cache the git conventions for later steps.

Before creating the branch, check whether a remote named origin exists (git remote get-url origin). If it does, check whether the branch exists on the remote (git ls-remote --heads origin <prefix>/<description>). If it exists, ask the developer how to proceed (reuse, rename, or delete). If no origin remote exists, skip the remote check and proceed.

4. Write test cases from acceptance criteria (TDD)

Read ./guidelines/tdd.md and follow its TDD process: extract criteria, discover test infrastructure, write tests (red phase), then implement (green phase).

If MCP available (check MCP Status in technical-context.md):

  • Use search_graph(label="Function", name_pattern=".*test.*") to find existing test patterns
  • Use get_code_snippet() to read relevant test examples

5. Functional commit

Once all tests are green, follow the functional commit conventions from the cached git guidelines (./guidelines/git.md).

6. Quality check

Read ./guidelines/quality-check.md and run all checks — linting, type-checking, security, and manual code review. Fix every issue found.

7. Final verification

  1. Run the tests again — confirm all are still green.
  2. If any regression was introduced, fix before proceeding.

8. Quality commit

Follow the quality commit conventions from the cached git guidelines (./guidelines/git.md).

9. Report

Tell the developer:

  • Branch name
  • What was implemented (one-line summary)
  • Test results (all green)
  • Linting/quality results (what was checked, what was fixed)
  • Commit SHAs (short form, e.g., a1b2c3d)

Throughout the workflow, follow the loaded global guidelines.

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.