plan-designer
Skill1 starsChanged 5 months agoMITView on GitHub
Use when creating, revising, or reviewing a coding implementation plan that must become bounded, decision-complete, testable, and safe for another engineer or coding agent to execute without guessing.
In this repository14 other files it gives its agents.
- AGENTS.mdAGENTS.md
- agent-skills-archive-orchestratorSkill
- call-r-lib-in-pythonSkill
- code-commenterSkill
- code-reviewerSkill
- end-user-xp-improverSkill
- fp-developerSkill
- lesson-documenterSkill
- preferred-workflowSkill
- release-preparerSkill
- scientific-prose-editorSkill
- simple-code-writerSkill
- spec-driven-developerSkill
- wayback-link-fallbackSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 4 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈4
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 25
Check that an implementer can identify the first three actions without making a design decision.
- Distinctive: line 28
Use the plan template when a detailed operational handoff is needed.
- Distinctive: line 46
Do not pad a small refactor to satisfy a template.
- Distinctive: line 52
Avoid new dependencies, public API changes, broad rewrites, generated-file edits, and opportunistic cleanup unless the request requires …
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Plan Designer
Turn an ambiguous coding request into the smallest operational plan that preserves existing behavior unless change is explicitly required.
Planning workflow
- Inspect repository instructions, current implementation, public contracts, tests, and relevant history before asking discoverable questions.
- Restate the requested outcome as observable behavior and identify what must remain unchanged.
- Resolve product preferences with the user; record technical assumptions that cannot be proven before implementation.
- Choose the smallest approach using existing platform, standard-library, framework, and dependency capabilities before proposing new machinery.
- Name likely edit targets, interfaces, data flow, concrete failure states, and migration or compatibility effects where applicable.
- Define focused tests, acceptance criteria, non-goals, stop conditions, and one risk label.
- Check that an implementer can identify the first three actions without making a design decision.
Use the plan template when a detailed operational handoff is needed. Use the quality checklist before returning a non-trivial plan. Consult ambiguity rewrites only when the draft contains vague goals, and the example plan only when the expected level of detail is unclear.
Required plan content
Adapt the presentation to task size, but include:
- Narrow task restatement and current repository understanding.
- Explicit assumptions and instructions to stop when they are false.
- Ordered implementation changes with likely files or bounded discovery targets.
- Public API, schema, I/O, compatibility, or migration effects—or an explicit statement that none are expected.
- Concrete test cases, commands, and observable acceptance criteria.
- Non-goals, stop conditions, handoff requirements, and exactly one risk label.
Do not pad a small refactor to satisfy a template. Do not hide uncertainty behind vague verbs such as “improve,” “robustify,” or “clean up.” Convert edge cases into explicit inputs or states.
Boundaries
- Avoid new dependencies, public API changes, broad rewrites, generated-file edits, and opportunistic cleanup unless the request requires them.
- Require the implementer to stop instead of improvising when repository reality conflicts with the plan.
- Reserve review for semantic correctness, scope, and risk—not only test passing.
- Label low risk for local changes, medium risk for multiple consumers or compatibility concerns, and high risk for public contracts, persistence, concurrency, security, payments, privacy, production configuration, or data loss.
Execution handoff
Include this instruction:
Implement exactly this plan. Do not broaden scope. If the plan conflicts with the codebase, stop and report the conflict instead of improvising.
Require the implementer to report files changed, tests run, deviations, and unresolved risks.
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
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.

