From Flagrare
Orchestrates ticket preparation: gathers context from Jira/Notion/Figma, grounds it in the codebase, then produces a TDD-first implementation plan.
How this skill is triggered — by the user, by Claude, or both
Slash command
/flagrare:work-prepThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Full preparation workflow: ticket context gathering followed by implementation planning.
Full preparation workflow: ticket context gathering followed by implementation planning.
This skill orchestrates two existing skills in sequence. It does not duplicate their logic.
SKU-123, ST-456, CORE-789)/flagrare:intakeCall /flagrare:intake with the ticket reference prefixed by [work-prep] (e.g., args: [work-prep] SKU-123). This prefix tells intake to skip its Step 6 next-step prompt and return control directly. This skill will:
/flagrare:codebase-explore — finding the files, utilities, and prior attempts the plan will touchsrc/x.ts or fork it?" — not abstract: "where should this live?")Wait for /flagrare:intake to complete before proceeding. The context brief must be finalized, codebase findings populated, and open questions resolved.
/flagrare:atdd-planOnce /flagrare:intake has produced a complete, codebase-grounded context brief, invoke /flagrare:atdd-plan. Pass the brief as opening context. This skill will:
/flagrare:codebase-explore pass (atdd-plan stays self-sufficient — intake's findings in the brief are additional input, not a substitute)After the plan is presented, issue an AskUserQuestion tool call — same interaction shape as intake's next-step prompt and plan-mode's accept tool. Do NOT phrase this as prose; that lets the turn end ambiguously.
Options:
/flagrare:intake and jump to planning. Context gaps turn into rework./flagrare:atdd-plan before clarifying questions are resolved./flagrare:intake already gathered. Pass the brief forward./flagrare:codebase-explore before Step 5./flagrare:codebase-explore runs twice in this flow — once in intake (scoped to inform questions) and once in atdd-plan (scoped to anchor the plan). atdd-plan stays standalone-callable; that requires it to do its own pass.AskUserQuestion prompt — same UX contract as plan-mode's accept tool. A button, not a typing prompt.The [work-prep] prefix in intake's args is the signal. When intake sees it, intake skips its next-step prompt and immediately invokes /flagrare:atdd-plan itself. This means:
[work-prep] prefix.If you are executing work-prep and intake finishes WITHOUT invoking atdd-plan (e.g., it returned with just the brief), you MUST invoke /flagrare:atdd-plan yourself before proceeding to Step 3.
/flagrare:work-prep [ticket ID or URL]
|
v
/flagrare:intake [work-prep] <- 1. read ticket + follow references in parallel
2. synthesise brief (no questions yet)
3. /flagrare:codebase-explore <- ground brief in code
4. ask codebase-informed clarifying questions
5. resolve, finalize brief
6. detect [work-prep] prefix → invoke atdd-plan directly
|
v
/flagrare:atdd-plan <- runs its own /flagrare:codebase-explore pass
(stays self-sufficient for standalone use),
then produces ATDD-first plan
|
v
[user reviews plan]
|
v
[implementation begins]
npx claudepluginhub flagrare/agent-skills --plugin flagrareProvides behavioral guidelines to reduce common LLM coding mistakes, focusing on simplicity, surgical changes, assumption surfacing, and verifiable success criteria.
Searches, retrieves, and installs Agent Skills from prompts.chat registry using MCP tools like search_skills and get_skill. Activates for finding skills, browsing catalogs, or extending Claude.
Creates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.