f166ec1a4ec7d3311226653ec020a7dab737e769
2 Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
FluxLuFFy
|
25ce2ca7bf |
fix: resolve 12 bugs across API, MCP, agent tools, web search, and context overflow (#674)
* fix: resolve 12 bugs across API, MCP, agent tools, web search, and context overflow API fixes: - Fix Gemini 400 error: delete 'store: false' field for Gemini endpoints (was globally injected, Gemini rejects unknown fields) - Fix session timeout 500 errors after ~25min: add 120s idle timeout on SSE stream readers in openaiShim and codexShim to detect dead connections and trigger withRetry reconnection - Fix context overflow 500 errors: add handler in errors.ts for 500 responses caused by oversized conversation context (too many tokens), surfacing user-friendly message with recovery actions instead of raw 'API Error: 500' Agent loop fix: - Fix premature task completion: detect continuation signals like 'so now I have to do it' in assistant text without tool calls and inject a meta nudge to force the agent to continue Web search improvements: - Increase result counts: Bing/Tavily/Exa/Firecrawl from 10→15, Mojeek/You/Jina from default→10 (explicit), max_uses 8→15 MCP fixes: - Reduce default tool timeout from ~27.8 hours to 5 minutes (tools no longer hang indefinitely on unresponsive servers) - Add retry logic (3 attempts) for tools/list fetch failures (prevents all MCP tools from silently disappearing on timeout) - Add abort signal check in URL elicitation retry loop - Improve MCP error messages with server and tool name context Agent tool fixes: - Fix SendMessage race condition: double-check task status before auto-resuming stopped agents to prevent duplicate registration - Fix auto-compact circuit breaker gap: when auto-compact fails 3+ consecutive times, proactively block oversized context BEFORE the API call instead of letting it 500. Clear message with recovery instructions (/new, /compact, rewind). Tests: 850 total, 0 failures (25 new bugfix tests) * fix: address all 4 review blockers + 6 additional issues from PR #674 Blockers (from Vasanthdev2004 review): 1. Continuation nudge infinite loop — no loop guard Added continuationNudgeCount to State, capped at MAX_CONTINUATION_NUDGES (3). Counter increments on each nudge, resets on tool execution (next_turn). 2. Continuation signal regexes too broad — high false-positive rate Tightened all patterns to require explicit action verbs. Added completion marker check (done/finished/completed/summary). Broad patterns only fire on messages <80 chars. 3. BUGFIXES.md in repo root — scope contamination Removed. PR description already contains this info. 4. AgentTool dump state cleanup is comment-only, not a bug fix Wrapped clearInvokedSkillsForAgent and clearDumpState in individual try/catch blocks so one failure doesn't prevent the other. Additional issues: 5+6. readWithTimeout ignores AbortSignal, timer leak on abort Added optional signal param to openaiStreamToAnthropic, codexStreamToAnthropic, collectCodexCompletedResponse, readSseEvents. Added abort listener that clears idle timer so AbortError surfaces cleanly instead of spurious idle timeout. 7. MCP error format change breaks consumers Reverted human-readable message to original errorDetails format. Moved server/tool context to telemetryMessage param only. 10. AgentTool test broken by comment change Updated test assertions to match new defensive cleanup text + try/catch. 12. Mojeek test regex dangerously broad Tightened to match searchParams.set('t', '10') specifically. 14. linkup.ts in providerCounts test — no result count field Removed from providers list (uses depth param, not result count). 15. Error message overlap between errors.ts and query.ts Prefixed errorDetails with 'Context overflow (500):' to distinguish. Tests: 851 pass, 0 fail --------- Co-authored-by: openclaude-bot <bot@openclaude.ai> Co-authored-by: Fix Bot <fix@openclaude.dev> |
||
|
FluxLuFFy
|
4ad6bc50c1 |
refactor: provider adapter system + 7 new search providers (bug-fixed) (#512)
* refactor: provider adapter system + 7 new search providers Architecture: - Each search backend is a small adapter implementing SearchProvider - 12 providers: custom, tavily, exa, you, jina, bing, mojeek, linkup, firecrawl, duckduckgo + native - WEB_SEARCH_PROVIDER controls selection: auto (fallback chain) or specific provider - Auth always in headers, never in query strings Bug fixes from review feedback: - Fix applyDomainFilters catch block: keep hits with malformed URLs on blocked_domains (can't confirm blocked), drop on allowed_domains (can't confirm allowed) - Add safeHostname() helper: safely extract hostname from URLs without throwing - Replace unsafe new URL(r.url).hostname in 7 providers with safeHostname() - Remove dead code: buildAllHeaders, buildAuthHeaders, parseExtraHeaders from types.ts - Fix WEB_PARMS typo: consistently use WEB_QUERY_PARAM everywhere - AbortSignal forwarded to fetch() in all 12 providers - DuckDuckGo: wrap dynamic import in try/catch for graceful error - Exa: remove double domain filtering (server-side already) - runSearch(): aggregate all provider errors instead of throwing only the last one - Retry logic: check numeric status code directly, retry 5xx/network, skip 4xx Test coverage (44 tests, all passing): - types.test.ts: safeHostname, normalizeHit, applyDomainFilters (20 tests) - index.test.ts: getProviderMode, getProviderChain, getAvailableProviders (13 tests) - custom.test.ts: extractHits flexible response parsing (11 tests) Co-authored-by: FluxLuFFy <195792511+FluxLuFFy@users.noreply.github.com> * security: add guardrails to custom search provider (Option B) - HTTPS-only by default (opt-out: WEB_CUSTOM_ALLOW_HTTP=true) - Private/localhost IPs blocked by default (opt-out: WEB_CUSTOM_ALLOW_PRIVATE=true) - Header allowlist: only known-safe headers allowed unless WEB_CUSTOM_ALLOW_ARBITRARY_HEADERS=true - Configurable timeout in seconds (WEB_CUSTOM_TIMEOUT_SEC, default 15) - Configurable POST body limit (WEB_CUSTOM_MAX_BODY_KB, default 300) - Removed max URL size restriction - Audit log warning on first custom search call - Updated .env.example and README_SEARCH_PROVIDERS.md with all new options * fix: remove custom provider from auto chain (Option 1) Remove customProvider from the auto fallback chain so it is only available when WEB_SEARCH_PROVIDER=custom is explicitly selected. Changes: - Remove customProvider from ALL_PROVIDERS array in providers/index.ts - Add 3 new tests verifying custom is excluded from auto chain - Update README_SEARCH_PROVIDERS.md: auto priority, mode table, note - Update .env.example: auto priority comment, custom mode annotation All 47 tests pass (44 existing + 3 new). Co-Authored-By: @Vasanthdev2004 * fix: address review blockers (routing, abort, config check, domain matching) 1. Native/Codex routing precedence in auto mode shouldUseAdapterProvider() now checks if native/first-party/vertex/foundry or Codex paths are available before falling back to adapter providers. Auto mode: native paths take precedence; adapter is fallback only. 2. AbortError stops provider chain immediately runSearch() now checks for AbortError/aborted signal before continuing the fallback chain. Cancelled searches don't create extra outbound requests. 3. Explicit provider mode fails fast on missing credentials runSearch() validates isConfigured() for explicit modes before attempting requests. Throws clear error: 'Search provider "X" is not configured.' 4. Domain filter exact-or-subdomain matching (fixes suffix collision) New hostMatchesDomain() helper: exact match or .subdomain match. badexample.com no longer matches example.com. 5. Tests: 56 pass (9 new) covering all 4 fixes Co-Authored-By: @Vasanthdev2004 --------- Co-authored-by: Claude Fix <fix@openclaude.local> Co-authored-by: FluxLuFFy <195792511+FluxLuFFy@users.noreply.github.com> Co-authored-by: bot <bot@openclaw.ai> |