From tac
Generates initial expertise.yaml for a codebase domain by exploring files and patterns. Bootstraps mental models for agent experts like database or websocket handling.
npx claudepluginhub melodic-software/claude-code-plugins --plugin tacThis skill is limited to using the following tools:
Generate initial expertise file by exploring the codebase and building a mental model.
Scaffolds agent expert in .claude/commands/experts/ with expertise.yaml, question.md, self-improve.md, plan.md, and plan-build-improve.md for Act-Learn-Reuse pattern. Invoke via /create-expert <domain>.
Transforms engineer expertise extracts into queryable skills with progressive disclosure, organizing by task, language, and pattern for AI agents to apply coding guidance efficiently.
Generates comprehensive 500-1000 line expert agent prompts for Claude Code with detailed patterns, code examples, and best practices. Triggers on 'spawn agent', 'create agent', 'generate expert' phrases.
Share bugs, ideas, or general feedback.
Generate initial expertise file by exploring the codebase and building a mental model.
$1: Domain name (required, e.g., "database", "websocket")$ARGUMENTS: Focus areas (optional, e.g., "connection-pool queries migrations")You are seeding an expertise file by exploring the codebase and extracting domain knowledge.
Extract:
$1 (required)If no domain provided, STOP and ask for domain name.
Check if expert directory exists using Glob:
Glob: .claude/commands/experts/{$1}/*
If not found:
Check if expertise already populated:
.claude/commands/experts/{domain}/expertise.yamlDelegate to the expertise-seeder agent with:
.claude/commands/experts/{domain}/expertise.yamlThe agent will:
After seeding completes:
## Expertise Seeded: {domain}
### Coverage
| Section | Entries |
| --- | --- |
| Core Implementation | X modules |
| Key Operations | X operations |
| Best Practices | X items |
| Known Issues | X items |
### File Stats
- Location: `.claude/commands/experts/{domain}/expertise.yaml`
- Lines: X/1000
- Valid YAML: Yes/No
### Next Steps
1. Review the generated expertise for accuracy
2. Run self-improve to validate: `/tac:improve-expertise {domain} false`
3. Test with a question: `/experts/{domain}/question "How does X work?"`
# Seed with default exploration
/tac:seed-expertise database
# Seed with focus areas
/tac:seed-expertise database connection-pool queries migrations
# Seed websocket expert
/tac:seed-expertise websocket events hooks frontend-integration
The expertise-seeder follows this approach:
Last Updated: 2025-12-15