Plan
My personal collection of AI skills
npx -y skills add abijith-suresh/skills --skill planAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 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
Presents a concrete implementation plan in chat based on a decisions summary. Use only when the user explicitly invokes the plan skill. Does not explore the codebase or ask clarifying questions. Do not trigger from ordinary planning requests.
SKILL.md
2.9 KB, as published. Nobody here has run it
Plan
What this skill does
Reads a decisions summary from the conversation and produces a concrete, human-readable implementation plan in chat. Does not re-explore the codebase — the decisions summary is the sole source of truth. Does not ask clarifying questions — assumes decisions are already made. Writes to a file only if explicitly asked.
Steps
1. Read the decisions summary
Find a decisions summary earlier in the conversation or in a project document. Look for it in:
- A decisions summary written in chat (e.g., by grill-me or the user)
- A handoff document (e.g.,
/tmp/handoff-*.md) - An existing project file with explicit decisions (e.g.,
IMPROVE.mdfrom the improve skill, a design doc, or a spec) - A conversation artifact (e.g., PRD, issue, or user-written notes)
Use the decisions summary as the sole source of truth for what to build and how. Do not re-derive decisions from scratch.
2. Present the plan in chat
Write the plan directly in chat using this structure:
# Plan: [Feature Name]
## Goal
[One sentence: what this achieves and why.]
## Steps
1. [Concrete step — name the file(s) to touch, the function or component
involved, and what changes. Not line-by-line, but specific enough that
a capable agent can act without ambiguity.]
2. ...
## Testing Notes
[Exact flows to verify, specific edge cases to check, and commands to run
— all derived from what actually changed in the steps above. No generic
advice, no boilerplate.]
3. Ask whether to save to PLAN.md
After presenting the plan in chat, ask:
Should I save this to
PLAN.md? Useful if you plan to run to-issues or continue in a new session.
Write to PLAN.md in the project root only if the user says yes.
If the user says no, or does not answer, do not write it.
4. Confirm
End with: "Ready to start implementing, or anything to adjust?"
Rules
- No codebase exploration — the decisions summary is the sole source of truth
- No clarifying questions — the decisions summary is assumed to be complete. If any decision is ambiguous, flag it in the plan and ask the user to resolve it
- Steps must name the files and components to be touched, not just describe actions at a high level
- Testing notes must be specific and derived from this plan — never generic
- Never write PLAN.md without asking first — chat is always the default
- Never run without a decisions summary present in the conversation or
a project document. The summary can come from any source — grill-me,
a user-written summary, a handoff document (
/tmp/handoff-*.md), anIMPROVE.md, a design doc, a spec, or any other structured artifact