Help us improve
Share bugs, ideas, or general feedback.
From Flagrare
Explores codebase to map conventions, reusable utilities, analogous features, and data flows for a planned change. Returns raw findings without producing a plan.
npx claudepluginhub flagrare/agent-skills --plugin flagrareHow this skill is triggered — by the user, by Claude, or both
Slash command
/flagrare:codebase-exploreThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Return raw exploration findings for a planned change. No plan, no tests, no implementation steps. Just facts about the codebase that a planning skill needs to make informed decisions.
Provides 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.
Fetches up-to-date documentation from Context7 for libraries and frameworks like React, Next.js, Prisma. Use for setup questions, API references, and code examples.
Share bugs, ideas, or general feedback.
Return raw exploration findings for a planned change. No plan, no tests, no implementation steps. Just facts about the codebase that a planning skill needs to make informed decisions.
/flagrare:atdd-plan before it writes acceptance testsReceive one of:
/flagrare:intake)Search for branches, PRs, and commits related to this work:
git branch -a | grep -i "<ticket-key-or-keyword>"
git log --all --oneline | grep -i "<ticket-key-or-keyword>"
gh pr list --search "<ticket-key-or-keyword>" --state all
If found, read them. They reveal: abandoned approaches, partial implementations, reviewer feedback, constraints discovered during prior attempts.
From the context brief or description, determine which directories and files the change will touch. Start broad, narrow fast:
The strongest predictor of how new code should look is how the codebase already solved a similar problem.
Search for:
Read 1-2 complete analogous files. These become the template.
For the feature area, identify:
| Convention | How to find it |
|---|---|
| File organization | ls the directories where changes will live |
| Naming patterns | Read 2-3 files in the same directory |
| Testing style | Read the closest existing test file |
| State management | How do sibling features manage state? |
| Error handling | How do sibling features handle failures? |
| i18n / messaging | What pattern do message definitions follow? |
| Imports / dependencies | What shared utilities are imported? |
Find utilities, helpers, shared components, and abstractions that the implementation should use instead of creating from scratch:
For each, note the import path and a one-line description of what it does.
.eslintrc, prettier.config, or equivalentReturn findings in this structure (no headers beyond these, no plan, no steps):
## Prior Attempts
[branches, PRs, commits found — or "none found"]
## Feature Area Map
[entry point → data flow → dependencies, with file paths]
## Analogous Features
[1-2 similar features with file paths and what makes them analogous]
## Conventions
[table of conventions observed in this area]
## Reusable Utilities
[list: import path + one-line description]
## Constraints
[lint rules, type system requirements, gating, project guidelines]
/flagrare:intake ← context brief produced
↓
/flagrare:atdd-plan
├── /flagrare:codebase-explore ← THIS SKILL (returns raw findings)
├── writes acceptance tests
├── names design patterns
├── SOLID audit
└── gap review