agentleFS
Flowerf19Flowerf19/agents-skills/implementation-planner/SKILL.md

implementation-planner

Skill3 starsChanged 4 days agoMITView on GitHub

Compare designs and plan code changes using the current project. Use when the user asks to plan a feature, choose an architecture, turn a spec into tasks, or plan a complex approved change. Planning does not give permission to implement.

7 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use 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.
  • 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

We split this skill into the rules it gives an agent

9 rules. 0 are common, 7 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 7

    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. Distinctive:

    Follow the affected flow from its starting point to its caller.

    line 16
  2. Distinctive:

    If a bug's cause is unclear, get a diagnosis first.

    line 17
  3. Distinctive:

    Ask about missing business rules or choices that change the approved scope.

    line 22
  4. Distinctive:

    Write the plan in plain language, following the "Plain writing" section in AGENTS.md (ISO 24495-1).

    line 26
  5. Very specific:

    Use the existing plan format.

    line 28
  6. Distinctive:

    Test commands, expected results, failure cases, and required environment, services, models, or files.

    line 36
  7. Distinctive:

    Add details only when relevant:

    line 38
  8. Distinctive:

    Check that each task has its prerequisites and a way to verify it.

    line 48
  9. Very specific:

    Use the project's format. Without an existing format, use status, created, and last_updated in YAML frontmatter.

    line 54

“Common” is measured: kept by at least 6 in 10 of the 5 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

Implementation planner

Recommend a design or produce a plan another developer can follow.

Plan only. Do not edit code, tests, configuration, dependencies, or unrelated guidance. For a discussion, answer in the conversation. Write a plan file only when requested or needed for an approved implementation.

Step 1: Frame the change

  1. State the goal, required behavior, constraints, and work outside the scope.
  2. Follow the affected flow from its starting point to its caller. Find the modules that create and use the data, where shared types are defined, and where dependencies are connected.
  3. If a bug's cause is unclear, get a diagnosis first.

Step 2: Resolve design questions

  1. Compare useful options for correctness, compatibility, cost, and testing. Recommend one. Keep proposals separate from accepted decisions.
  2. Ask about missing business rules or choices that change the approved scope. Keep blocking questions open until answered.

Step 3: Write the needed plan

Write the plan in plain language, following the "Plain writing" section in AGENTS.md (ISO 24495-1). Put the goal first, number the steps, and explain each new term once.

Use the existing plan format. If none exists, put an approved plan file in .agents/plans/<slug>.md. Create only the folder needed for that file.

Include:

  • Goal, approved scope, and checks that show success.
  • Accepted decisions and their sources; open questions separately.
  • Small tasks, affected files or modules, and prerequisites. Follow dependencies, not task-number order.
  • Required checks before starting the next phase or declaring the work complete.
  • Test commands, expected results, failure cases, and required environment, services, models, or files. Separate unit tests, integration checks, and checks of the actual product flow. State what remains unverified if a required check cannot run.

Add details only when relevant:

  • Interface changes: inputs, outputs, errors, callers, and behavior that must stay the same.
  • Refactors: public imports, API payloads, event order, and stored-data compatibility.
  • State changes: cancellation, concurrent updates, restart, migration, and rollback.
  • UI changes: routes, shared components, and current design tokens.
  • Benchmarks or classifiers: agreed metrics, representative cases, and counterexamples. Do not invent score thresholds.

Step 4: Check and hand off

Check that each task has its prerequisites and a way to verify it. A plan with blocking questions is not ready to execute.

Return the recommendation or plan path, supporting evidence, open questions, and whether implementation is approved.

Plan status

Use the project's format. Without an existing format, use status, created, and last_updated in YAML frontmatter.

StatusMeaning
draftProposed work; execution is not yet approved
in-progressApproved execution has started
doneApproved tasks are implemented and verified
abandonedReplaced or stopped; record the reason

Preserve task IDs and completed history. Update last_updated when the plan changes. Record scope changes explicitly. An accepted design can appear in a draft plan; it does not grant execution permission. Deleting a plan needs its own permission.

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.

Reviews can't be read right now.

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.