feat: wire dead-code command to /v1/analysis/dead-code endpoint

[jonathanpopham]

Problem

The dead-code command currently calls /v1/graphs/supermodel, gets a generic graph, and does naive edge-counting in Go (finds Function nodes with no incoming calls edges). This misses the API's real dead code analysis which includes:

• Multi-phase reachability (root file detection, entry point identification, BFS traversal)
• Transitive dead code propagation
• Confidence levels (high/medium/low)
• Line numbers and reasons
• Framework-aware entry point detection (decorators, lifecycle methods)
• Optional LLM verification to reduce false positives

The API already exposes all of this at POST /v1/analysis/dead-code.

Solution

Wire the CLI's dead-code command to the dedicated endpoint instead of the generic graph endpoint:

1. Add DeadCode() method to api.Client that POSTs to /v1/analysis/dead-code with async polling (same pattern as Analyze())
2. Add response types for the dead code envelope (DeadCodeResult, DeadCodeCandidate, etc.)
3. Rewrite internal/deadcode/handler.go to call the new endpoint and display the rich results
4. Surface confidence, line numbers, and reasons in CLI output
5. Support --min-confidence and --limit query params from the API

Expected output (after)

FILE                        LINE  FUNCTION              CONFIDENCE  REASON
src/utils/idGenerator.ts    8     generateSequentialId  high        No callers found in codebase
src/helpers/deprecated.ts   42    oldParser             medium      Only called from dead code

References

• API spec: POST /v1/analysis/dead-code in openapi.yaml
• Current CLI handler: internal/deadcode/handler.go (naive edge counting)
• Async polling pattern: internal/api/client.go Analyze() method

[coderabbitai]

🔗 Related PRs

#15 - fix: use correct API endpoint /v1/graphs/supermodel [closed]
#16 - fix: handle async 202 job polling for analyze [closed]

---

📝 Issue Planner

_{Check the box below or use the @coderabbitai plan command to generate an implementation plan and prompts that you can use with your favorite coding assistant.}

• Create Plan

---

🧪 Issue enrichment is currently in open beta.

To disable automatic issue enrichment, add the following to your .coderabbit.yaml:

issue_enrichment:
  auto_enrich:
    enabled: false

💬 Have feedback or questions? Drop into our discord!