From agentic-council
Codebase architecture review — find shallow modules, tight coupling, and deepening opportunities
How this command is triggered — by the user, by Claude, or both
Slash command
/agentic-council:deepen [PATH or SCOPE...]The summary Claude sees in its command listing — used to decide when to auto-load this command
# /deepen — Architecture Review Explore the codebase for structural improvement opportunities. Finds shallow modules, unnecessary abstractions, and tight coupling — then presents concrete deepening candidates. ## Usage ## Instructions When the user runs `/deepen`, follow these steps: ### 1. Scope Resolution If a path or scope is provided in `$ARGUMENTS`, focus there. Otherwise, scan the project root. Detect workspace: `git rev-parse --show-toplevel` or `$PWD`. Read project structure, tech stack markers (`package.json`, `pyproject.toml`, etc.), and `CLAUDE.md` for conventions. ### ...
Explore the codebase for structural improvement opportunities. Finds shallow modules, unnecessary abstractions, and tight coupling — then presents concrete deepening candidates.
/deepen
/deepen src/api
/deepen "the auth system"
When the user runs /deepen, follow these steps:
If a path or scope is provided in $ARGUMENTS, focus there. Otherwise, scan the project root.
Detect workspace: git rev-parse --show-toplevel or $PWD. Read project structure, tech stack markers (package.json, pyproject.toml, etc.), and CLAUDE.md for conventions.
Map the codebase structure within scope:
Use Explore agents for large scopes — up to 3 in parallel, each focused on a different area.
Load commands/references/deepening-criteria.md for the evaluation rubric.
For each module or area, assess:
For each finding, present:
### [Module/Area Name]
**Signal:** [which criterion triggered — e.g., "Shallow module", "Tight coupling"]
**Evidence:** [specific files, exports, import counts]
**Suggestion:** [concrete action — merge these modules, simplify this interface, hide this behind X]
**Impact:** [High/Medium/Low — how much complexity this eliminates]
Rank opportunities by impact-to-effort ratio. Recommend a starting point — the change that unlocks the most clarity for the least disruption.
If findings are extensive, suggest: "Run /council on the top candidates for deeper architectural deliberation."
npx claudepluginhub dtsong/agentic-council --plugin agentic-council/deepenRefactors a shallow module into a deeper design with a smaller interface, preserving all behavior and passing tests before and after.