- Home
- Skills
- Catlog22
- Claude Code Workflow
- Req Plan With File
req-plan-with-file_skill
- TypeScript
1.3k
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 catlog22/claude-code-workflow --skill req-plan-with-file- SKILL.md37.2 KB
Overview
This skill provides requirement-level progressive roadmap planning with JSONL output and explicit convergence criteria. It decomposes a requirement into either layered MVP→iterations or a topologically-sorted task sequence, each item accompanied by testable verification steps and a business definition of done. The output is designed for automated consumption by CLI orchestration and downstream planning tools.
How this skill works
The tool parses a freeform requirement, assesses uncertainty across five dimensions, and selects a decomposition strategy (progressive, direct, or auto). It optionally explores the codebase for relevant modules and constraints. Finally it generates validated JSONL records (layers or tasks) with convergence criteria, performs schema and quality checks, and writes a human-readable roadmap.md alongside roadmap.jsonl for review and iteration.
When to use it
- When you need a testable, machine-readable development roadmap from a single requirement statement.
- When uncertainty is high and you want an MVP-first, iterative decomposition (progressive mode).
- When the requirement is clear and you prefer an ordered task sequence with explicit dependencies (direct mode).
- When you want outputs compatible with automated CLI workflows and multi-agent orchestration.
- When you need convergence criteria for gated releases, demos, or automated verification.
Best practices
- Provide a concise requirement that includes goals, constraints, and stakeholders to improve decomposition quality.
- Use progressive mode for research-heavy or ambiguous problems; use direct mode for well-defined engineering tasks.
- Include project context files (package.json, project-tech.json) in the repo to enable deeper codebase exploration.
- Review and iterate up to the allowed feedback rounds; use -y only when confident in auto mode recommendations.
- Ensure convergence criteria are executable commands or verifiable acceptance tests to enable automation.
Example use cases
- Generate an MVP→iteration roadmap for a new OAuth+2FA authentication feature with testable sign-in flows.
- Produce a topologically-sorted task list to refactor a payment module with explicit dependency order and verifications.
- Auto-select a decomposition strategy and produce JSONL for a caching layer, skipping interactive confirmations.
- Explore an existing codebase to locate integration points and output a validated roadmap.jsonl for automation.
- Export a roadmap.md and roadmap.jsonl to feed into a lite-plan executor or issue generator.
FAQ
The skill writes validated roadmap.jsonl (records per line) and a human-friendly roadmap.md. It also stores strategy-assessment.json and optional exploration-codebase.json.
How is strategy selected in auto mode?
It assesses five uncertainty dimensions (scope clarity, technical risk, dependency unknowns, domain familiarity, requirement stability). A rule-based threshold or explicit request decides progressive vs direct; -y auto-confirms.