From codebase-scribe
Mechanical stub creator. Receives an approved topic list from the orchestrator and creates stub files. Does NOT scan, propose, or make decisions.
How this skill is triggered — by the user, by Claude, or both
Slash command
/codebase-scribe:scribe-discoverThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
1. **STUBS ONLY.** Create files with YAML frontmatter + placeholder skeleton. Zero real content.
docs/agents/.You are a FILE CREATOR. You receive a list of approved topics with their watch_paths and migration info. You create one stub file per topic. That's it.
The orchestrator tells you exactly what to create. Each topic has:
backend-architecture)Backend Architecture)["cmd/", "pkg/", "business/"])"AGENTS.md") or nullCreate each file at docs/agents/<name>.md with this EXACT format:
---
scribe:
scan: null
freshness: 0
human_input: 0
completeness: 0
inferred_sections: []
watch_paths: ["from/", "orchestrator/"]
stale_flags: []
migration_source: "AGENTS.md"
migration_sections:
- "## Section Name"
---
# Topic Title
> What this doc covers and what it doesn't.
## Key Entry Points
*Stub — will be populated by the draft skill.*
## Patterns & Conventions
*Stub — will be populated by the draft skill.*
## Gotchas
*Stub — will be populated by the draft skill.*
## Dependencies & Context
*Stub — will be populated by the draft skill.*
## Links
*Stub — will be populated by the draft skill.*
migration_source/migration_sections ONLY if the orchestrator specified them for this topicwatch_paths to what the orchestrator providedCreate docs/agents/STATUS.md showing all topics as stubs with 0% scores.
If no AGENTS.md exists, create a hub with <!-- scribe:managed --> as the first line, followed by: Project Identity, Quick Reference, Architecture at a Glance, Documentation links, Conventions placeholder. In the "Architecture at a Glance" section include a line: > For the full architecture index, see [ARCHITECTURE.md](ARCHITECTURE.md). If AGENTS.md already exists, do NOT touch it.
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.
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.
npx claudepluginhub tommasobagassi/codebase-scribe --plugin codebase-scribe