plan_skill
- HTML
2
GitHub Stars
1
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill benredmond/apex --skill plan- SKILL.md13.4 KB
Overview
This skill transforms research outputs into a rigorous, evidence-backed technical architecture during the ARCHITECT phase. It interactively produces five mandatory artifacts (Design Rationale & Evidence, Tree of Thought, Chain of Draft, YAGNI Declaration, Pattern Selection) and enforces contract-aligned decisions. The process is iterative, verification-driven, and stops for human judgment when needed.
How this skill works
The skill reads the task file and verifies the research phase and task-contract as the authoritative scope. It spawns parallel verification agents to extend research intelligence, map system flows, surface git history, predict failures, and enumerate risks. After a human checkpoint it generates three distinct architecture options, iterates through three drafts, and emits the five required artifacts plus an architecture decision record and builder handoff.
When to use it
- You have completed research and need a production-ready implementation plan.
- You must translate findings and acceptance criteria into concrete files, patterns, and validation steps.
- You need multiple alternative architectures evaluated with evidence and risk ratings.
- You must lock scope with a YAGNI declaration to avoid feature creep.
- You require pattern choices tied explicitly to research-sourced confidence.
Best practices
- Read all referenced research files fully before any design work.
- Treat the task-contract as the single source of truth; stop if it is missing or ambiguous.
- Be interactive: require explicit user confirmation at checkpoints before advancing.
- Document evidence for every decision: code refs, commits, or research citations.
- Avoid fabricating patterns — only use patterns present in the research context.
Example use cases
- Convert a research brief into an implementation sequence with concrete files to change and create.
- Compare three fundamentally different approaches (monolith refactor, service split, adapter layer) and pick a winner with risks and complexity scored.
- Produce a YAGNI-bound plan for a minimum viable production rollout while noting deferred enhancements.
- Create an architecture decision record for builders with automated and manual validation gates.
- Validate that acceptance criteria are covered or append formal amendments to the contract.
FAQ
Provide the task identifier. The skill will prompt for it and points to the task directory; it cannot proceed without loading the task file and verifying the research phase.
Can the skill propose patterns not present in research?
No. Pattern claims must come from the research context. If a needed pattern is missing, the skill documents the gap and provides a verified workaround.