Commit Graph
16 Commits
Author SHA1 Message Date
Cokky Turnip e3e3e235f6 fix(gemini): fill empty tool schemas after $ref strip
Vertex rejects orphan {} left when $ref/$defs are removed from function declarations. Promote empty nodes to object+reason placeholder in addPlaceholders.
2026-07-29 19:31:15 +07:00
46e6c01a01 fix(claude): reconcile max_tokens vs thinking budget and lift per-model ceiling (#2381)
On the translated OpenAI->Claude path, adjustMaxTokens capped max_tokens
before applyThinking set thinking.budget_tokens, so max-effort budget
(128000) could exceed a 64k-clamped max_tokens -> Anthropic 400.
prepareClaudeRequest now reconciles after the budget is known: prefer
raising max_tokens, only shrink budget when it meets/exceeds the ceiling.

Also lift the global 64000 cap: the ceiling is now the model's real
maxOutput, so high-output models (fable/mythos, opus-4.8/sonnet-4.6) get
their full budget. adjustMaxTokens gains an optional ceiling arg (default
unchanged, callers untouched); openai-to-claude passes the model maxOutput.

Native Claude Code passthrough is unaffected.

Co-Authored-By: Claude <noreply@anthropic.com>
Co-authored-by: Cursor <cursoragent@cursor.com>
2026-07-05 17:38:16 +07:00
decoluaandCursor cd557a2552 fix(claude): drop foreign thinking signatures in passthrough
Combo mixes models, so non-Claude thinking signatures leak into
conversation history. Native passthrough forwarded them verbatim and
Anthropic rejected the request. Validate signatures and drop invalid
thinking blocks, re-inserting a placeholder when tool_use requires one.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-07-03 15:06:19 +07:00
RexandCursor 9e3866658a fix(alicode): preserve cache_control for DashScope providers (#2069)
Opt-in quirk preserveCacheControl keeps cache_control on content blocks
for alicode/alicode-intl, enabling DashScope prompt caching. signature
is always stripped; all other providers unchanged.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-29 15:29:49 +07:00
Yudhistira-OfficialandCursor 319caa2d7b fix(antigravity): strip 'deprecated' from tool schemas before Gemini
Gemini rejects the non-standard 'deprecated' keyword in nested tool
schemas with INVALID_ARGUMENT (400). Add it to UNSUPPORTED_SCHEMA_CONSTRAINTS
alongside 'optional' so it gets stripped during translation.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-29 15:28:02 +07:00
iletaiandCursor 3d20a4ccd2 fix(antigravity): strip deprecated/readOnly/writeOnly from tool schemas
Gemini/Antigravity generateContent rejects the JSON Schema annotation
keywords deprecated, readOnly, writeOnly with a 400 INVALID_ARGUMENT.
MCP tool schemas (e.g. Claude Code) commonly set deprecated:true, making
every request with such a tool fail. Add them to
UNSUPPORTED_SCHEMA_CONSTRAINTS so cleanJSONSchemaForAntigravity removes
them recursively before the request is sent.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-29 15:12:17 +07:00
Mink NguyenandCursor c4f80d30d8 fix provider thinking compatibility
- claude: handle DeepSeek thinking blocks defensively, unsigned placeholder; fix kept-vs-seen thinking detection
- gemini: clamp unsupported max/xhigh thinking levels to high
- testUtils: probe Cloud Code Assist for gemini-cli/antigravity with 401 refresh retry
- tests: add translator regression coverage

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-26 10:12:48 +07:00
decoluaandCursor 45240c19e5 fix(translator): normalize tools to Anthropic-native shape for non-Anthropic providers
Strip `type` field and fold `function.{name,description,parameters}` into
top-level {name, description, input_schema} before forwarding to Claude-format
endpoints. MiniMax (and other Anthropic-compatible providers) reject tools
carrying a `type` field with error code 2013 ("invalid tool type").

Refs #1939

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-21 16:26:17 +07:00
decolua 401d93bd5c fix(claude haiku): update handling of unsupported adaptive thinking and output_config.effort 2026-06-20 16:29:15 +07:00
mustafabozkayaandCursor f6c2f7cae3 fix(gemini): preserve 'pattern' in antigravity tool schema translation
Remove "pattern" from UNSUPPORTED_SCHEMA_CONSTRAINTS so glob/grep tool
schemas keep the required "pattern" property for the antigravity provider.

Fixes decolua/9router#1368

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-20 15:31:22 +07:00
decolua 637dd7ae66 feat(ponytail): introduce "Ponytail" feature for minimalistic code generation 2026-06-20 15:08:11 +07:00
Khanh LeandCursor db9ec3af61 fix(antigravity): strip optional from tool schemas before Gemini
Closes #1537, #1841

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-17 09:46:41 +07:00
decolua d03f9fb823 Enhance configuration and model capabilities 2026-06-16 23:32:28 +07:00
decolua b282f05549 Refactor 2026-06-15 18:18:04 +07:00
decoluaandCursor aba4c45da6 fix(translator): ESM-safe registry + tool-id pairing + responses max_tokens; add real-creds tests
- translator/index.js: replace require() with static side-effect imports (ESM-safe),
  lazy-init registry maps to survive circular import order
- openai-responses->openai: map max_output_tokens -> max_tokens (avoid leaking field upstream)
- gemini/antigravity -> openai: derive deterministic tool_call id from name so
  functionCall/functionResponse pair correctly (fixes provider tool-pairing 400s)
- add offline unit tests (finish-reason, usage, session-manager, ollama malformed args, const guard)
- add real-creds integration tests (provider-cases + all-formats matrix: 6 inbound formats x 4 scenarios)

Includes co-located provider registry refactor (pricing/capabilities/media providers) and sessionManager updates.

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-15 11:38:43 +07:00
decoluaandCursor d3f61aac2f refactor(open-sse): translator DRY + schema enums, bug fixes, dead code cleanup
- Bug B1-B7: media UI m.kind||m.type, serviceKinds, gemini mediaPriority, schema kind, models/info lookup by kind
- Dead code D1-D6: safeParseJSON, drop PROVIDER_ENDPOINTS, orphan fetcher, GITHUB_CONFIG derive, getProviderConfig internal, legacy kiro file
- Translator concerns: toOpenAIUsage, toOpenAIFinish (gemini/kiro/ollama + fix kiro tool finish), thinking effort maps
- Reorg helpers/ → concerns/ (logic) + formats/ (per-format) + schema/ (pure enums: roles/blocks/finishReasons/defaults)
- Wire ~280 hardcoded role/block/finish/default literals to schema enums across 20+ files
- collapseTextParts + extractTextContent dedup
- Normalize translator fn names to openaiToXRequest / xToOpenAIResponse
- Golden tests lock behavior; 0 regression (byte-for-byte providers/alias, 26=26 known fails)

Co-authored-by: Cursor <cursoragent@cursor.com>
2026-06-14 18:49:38 +07:00