diet-evaluate-removal

Diet Removal Evaluation: $ARGUMENTS

Deep-dive evaluation of whether $ARGUMENTS is worth removing, using a 5-phase data-driven framework. This command picks up where uzomuzo diet leaves off -- diet gives you the priority ranking, this gives you the removal verdict.

When to use: After uzomuzo diet ranks a dependency and you need to decide whether to invest the effort. Works for any difficulty level -- trivial deps get a quick confirmation, hard deps get a concrete migration plan.

Relationship to other skills:

• uzomuzo diet tells you ranking and difficulty (automated, all dependencies at once)
• /diet-evaluate-removal tells you whether it's worth removing and how (LLM-powered, one dependency at a time)
• /diet-assess-risk tells you how dangerous it is to keep (LLM-powered, security focus)
• /diet-remove executes the removal (LLM-powered, implementation)

---

Phase 1: Ingest Diet Data

Extract the target dependency's diet JSON entry. Prefer existing JSON over re-running diet.

Option A -- existing diet JSON (preferred):

jq '.dependencies[] | select(.name == "$ARGUMENTS")' diet.json

Option B -- run diet now:

uzomuzo diet --sbom bom.json --source . --format json | jq '.dependencies[] | select(.name == "$ARGUMENTS")'

Check that coupling was actually measured

difficulty and coupling_effort are only meaningful when diet matched imports in the source tree. When source analysis is skipped, fails, or finds no import of any dependency (typically because --source points at the wrong directory; diet logs no imports matched any dependency), diet gives every entry without a scope import_file_count = call_site_count = 0, coupling_effort = 0, difficulty = "trivial", and is_unused = false (tool and runtime entries get a placeholder api_breadth = 1 and show easy). Here "trivial" means "not measured", not "easy to remove".

Detect it across the whole JSON, not from the one entry: