- mitm-handler-base: hookBufferStart returns InterceptedRequest with
sanitized headers, extractSourceModel parses body.model, writeError
emits JSON with sanitized message (Hard Rule #12).
- mitm-handler-<id>: nine per-agent happy-path tests (antigravity,
kiro, copilot, codex, cursor, zed, claude-code, open-code) exercise
full intercept() with mocked fetch via _mitmHandlerHarness.ts;
asserts mapped model is forwarded and AgentBridge correlation
headers are present. Trae test confirms intercept() rejects with a
structured error.
- mitm-targets-resolve: ALL_TARGETS contains 9 entries; resolveTarget
is case-insensitive and returns null for unknown hosts.
- mitm-targets-route: bypass > target > passthrough precedence
validated against representative hostnames.
- mitm-detection: DETECTORS covers every AgentId; detectAgent never
throws and returns DetectionResult-shaped objects for all probes.
Adds 5 new settings route handlers for CLIs introduced by plan 14 that
declare configType:"custom" and need automated config file persistence:
forge (~/.forge/config.toml), jcode (~/.jcode/config.json),
deepseek-tui (~/.config/deepseek-tui/config.toml),
smelt (~/.smelt/config.json), pi (~/.pi/config.json).
Also registers the 5 tools in cliRuntime.ts path table so
getCliPrimaryConfigPath() resolves their config paths correctly.
Each handler follows the established pattern: requireCliToolsAuth guard on
every exported method, Zod body validation on POST, buildErrorBody/
sanitizeErrorMessage on all error paths (Hard Rule #12), fs/promises only
(no exec/spawn — Hard Rule #13), saveCliToolLastConfigured on success.
Integration tests: 7 subtests per handler (401 without auth, 200 GET,
400 missing-baseUrl, 400 missing-model, 200 POST writes file, 200 DELETE,
error sanitization + no exec/spawn static audit).
- compliance-tab-actor-filter.test.tsx: 4 tests verifying actor input
presence, initial fetch has no actor param, re-fetch with actor=X
after typing, and clearFilters does not throw.
- compression-log-namespace.test.tsx: 4 tests verifying that the logs
namespace is used (not settings), no _MISSING_ sentinels, and all 4
required keys are covered.
Add tests for default branch paths (model/prompt/stream falsy), completions
params branch, search with model set, web.fetch depth=0, reversed markers
in parseImprovedContent. All new production files reach 100% branch coverage.
16 tests covering all 5 endpoints:
- GET /agent-skills: 42 total, category/area filters, invalid category → 400
- GET /agent-skills/[id]: found api+cli skills, unknown → 404
- GET /agent-skills/[id]/raw: unknown → 404, valid → 200/502/500 (network-tolerant)
- GET /agent-skills/coverage: SkillCoverage shape (api.total=22, cli.total=20)
- POST /generate: no auth → 401/403, invalid body → 400, no generator → 503
Hard Rule #12 verified explicitly in every error case: error.message must
not match /\bat \/|\bat file:\/\// (no stack trace exposure). Final test
collects all error paths and asserts sanitization in a single sweep.
Adds 6 tests for executeListCapabilities (§3.7 shape, 42-skill IDs in markdown,
coverage bounds, ISO datetime, handler registration) and 5 tests for the Agent
Card route (6 skills total, list-capabilities presence + tags + examples, all
original 5 skills preserved). All 11 pass.
- Extracted Compression Preview from PlaygroundMode lines 506-584 into standalone accordion
- Wrapped in Collapsible with lazy-render (D7): content only mounts after first open
- Accepts inputContent via prop (lifted from TranslateTab in F9) or shows empty-state hint
- POST /api/compression/preview unchanged; error path sanitized (no stack-trace leak)
- 33 Vitest tests covering smoke, lazy-render guard, mode select, fetch dispatch, result grid, error path
Three new test files:
- sidebar-monitoring-reorg.test.ts: asserts monitoring has 4 children (activity item + logs/audit/system groups), no costs-parameters group, no logs-activity in items
- sidebar-costs-section.test.ts: asserts costs section exists with 4 items in correct order, costs removed from analytics, costs positioned between analytics and monitoring
- sidebar-back-compat.test.ts: asserts activity added + logs-activity preserved in HIDEABLE_SIDEBAR_ITEM_IDS, admin preset shows activity and hides logs-activity (B30)
- MONITORING_ITEMS reduced to single `activity` item at `/dashboard/activity`
- New LOGS_GROUP (logs/logs-proxy/logs-console) extracted from flat monitoring items
- New SYSTEM_GROUP (health/runtime) extracted from flat monitoring items
- AUDIT_GROUP preserved unchanged
- Monitoring section children: [...MONITORING_ITEMS, LOGS_GROUP, AUDIT_GROUP, SYSTEM_GROUP]
- COSTS_PARAMS_GROUP removed from monitoring section (items migrate to COSTS_ITEMS)
- `activity` added to HIDEABLE_SIDEBAR_ITEM_IDS; `logs-activity` preserved for back-compat (B11)
- Updated existing sidebar-visibility.test.ts to match new monitoring item structure
