[hsuchifeng]

Summary

Follow-up to #3 (digest output in English, no per-user intervention / token rankings). The printed digest on main is already English and no longer prints the ranking lines, but summarizeInterventions / summarizeConversation still computed and returned per-user ranked arrays that nothing used. This PR:

• removes the dead ranked fields so individual rankings cannot quietly come back through the summary API;
• adds a real-CLI e2e test (src/__tests__/digest-output-e2e.test.ts): a git provider team with two members' reported stats on teamai-reports, then teamai digest must contain no CJK characters, print the English autonomy / token sections with team-wide totals, and never print a member name or the word "rank";
• updates docs/usage-guide.md / docs/usage-guide.zh-CN.md, which still described per-person rankings.

Type of Change

• Bug fix (non-breaking change that fixes an issue)
• New feature (non-breaking change that adds functionality)
• Breaking change (fix or feature causing existing behavior to change)
• Documentation only
• Refactor / internal cleanup

Test Plan

• npx tsc --noEmit passes
• npm run lint passes
• npx vitest run passes (full unit suite)
• Real CLI e2e: npm run build && npx vitest run --config vitest.e2e.config.ts src/__tests__/digest-output-e2e.test.ts src/__tests__/intervention-e2e.test.ts src/__tests__/conversation-token-e2e.test.ts → 7/7 passed. Sample output from the run:

🤖 Session Autonomy (Human Intervention):
  Team avg: 1.43 interventions/session (14 sessions, 20 interventions)
  Breakdown: interrupt 14 · reject 2 · correction 4

💬 Lifetime Conversation & Token Usage:
  Lifetime human prompts: 130
  Lifetime tokens: 12.0K (input 6.0K · output 6.0K · cache read 0 · cache write 0)

• Added/updated tests: new digest-output-e2e.test.ts; digest-interventions, conversation-token-metrics, conversation-token-e2e, intervention-e2e no longer assert on ranked

Related Issues

Covers the scenario from #3.

Notes for Reviewers

ranked was only read by tests; the summary interfaces lose that field. The teamai.yaml wording in skill-data/ does not mention rankings, so only the usage guides needed a docs update.

---

Generated by Claude Code

[github-actions]

Code Erosion Report

Informational — never blocks the merge. scb-check==0.2.0 SlopCodeBench metrics; method & caveats in docs/ci-code-erosion.md.

Metric	Value	Human band	Agent band	Reading
Verbosity*	0.073	0.15	0.33	human
Erosion	0.685	0.31	0.68	agent
Cognitive erosion	0.867	–	–	–

Scanned 304 files / 60608 SLOC · high-CC functions 420/4834

* On TypeScript, verbosity is partial and the bands are Python-calibrated.

Rule hits (TS verbosity layer)

Standalone ast-grep, separate from the number above.

Rule	Hits
unnecessary-else-after-return	32
empty-catch-block	5
redundant-ternary-same	1

38 distinct lines flagged across src/.