(janitor/test-prune): prune tautological and duplicate KiloClaw plugin tests

[kilo-code-bot]

Summary

Prunes three KiloClaw plugin tests that add maintenance cost without protecting meaningful behavior. Test-only change; no production code touched.

Removed tests and surviving coverage

1. services/kiloclaw/plugins/kiloclaw-morning-briefing/src/index.web-search-query.test.ts (deleted, 3 cases)

   • buildBriefingWebSearchQuery is a single branchless template literal (\latest news and updates on ${topic} from the last 24 hours`). All three cases hardcode that exact literal, so they reproduce the production body and compare it with itself. The only way they fail is a cosmetic wording edit, not a behavioral regression. Surviving coverage: the function is exercised through collectWebSearchinindex.lifecycle.test.ts`; the per-input interpolation has no branch/edge case to protect.

2. services/kiloclaw/plugins/kilo-chat/src/client.test.ts (removed 1 case: createMessage posts to /_kilo/kilo-chat/send and returns messageId)

   • Strict subset of the surviving posts to controller /_kilo/kilo-chat/send with gateway token and conversation id test: same mocked response, same URL, method, and body assertions. The surviving test additionally asserts authorization and content-type headers, so it strictly dominates. No distinct failure mode is lost.

3. services/kiloclaw/plugins/kilo-chat/src/channel.test.ts (removed 1 case: routes send actions before constructing the shared action client)

   • Implementation-detail test that read channel.ts from disk and asserted indexOf ordering of source fragments. It inspects source layout, not runtime behavior, and would fail on a harmless refactor (e.g. extracting the send branch into a helper) while passing on a real logic bug elsewhere. The actual send / upload-file runtime outcomes are already covered by the two base64 buffer behavioral tests in the same file, which drive handleAction end-to-end through attachment upload + send.

Verification

• pnpm --filter @kiloclaw/kilo-chat test → 199 passed (was 201)
• pnpm --filter @kiloclaw/kiloclaw-morning-briefing test → 239 passed across 17 files (was 242/18)
• pnpm --filter @kiloclaw/kiloclaw-morning-briefing typecheck → clean
• oxfmt on changed files → clean; git diff --check → clean
• These packages are excluded from the repo lint scope (scripts/lint-all.sh skips services/kiloclaw/plugins/*).

Postgres/Redis-backed suites were not run (not available in the sandbox); all removed tests are pure unit tests with no DB dependency.

Why these could not detect an additional regression

Each removed test is either a self-mirror of a one-expression function, a strict duplicate of a stronger surviving assertion, or a source-text ordering check whose behavior is already asserted directly. None guards a branch, edge case, or contract absent from the surviving coverage.

[kilo-code-bot]

Code Review Summary

Status: No Issues Found | Recommendation: Merge

Executive Summary

Test-only prune of three tautological or duplicate KiloClaw plugin tests; surviving coverage still exercises createMessage, send/upload-file runtime, and web-search collection.

Files Reviewed (3 files)

• services/kiloclaw/plugins/kilo-chat/src/channel.test.ts
• services/kiloclaw/plugins/kilo-chat/src/client.test.ts
• services/kiloclaw/plugins/kiloclaw-morning-briefing/src/index.web-search-query.test.ts

---

_{Reviewed by grok-4.6 · Input: 53.3K · Output: 5.9K · Cached: 228.6K}

_{Review guidance: REVIEW.md from base branch main}