Commit Graph

475 Commits

Author SHA1 Message Date
Luis Pater
bc61bf36b2 Merge pull request #214 from anilcancakir/fix/github-copilot-model-alias-suffix
fix(auth): strip model suffix in GitHub Copilot executor before upstream call
2026-02-11 21:06:58 +08:00
Anilcan Cakir
bcd2208b51 fix(auth): strip model suffix in GitHub Copilot executor before upstream call
GitHub Copilot API rejects model names with suffixes (e.g. claude-opus-4.6(medium)).
The OAuthModelAlias resolution correctly maps aliases like 'opus(medium)' to
'claude-opus-4.6(medium)' preserving the suffix, but the executor must strip the
suffix before sending to the upstream API since Copilot only accepts bare model names.

Update normalizeModel in github_copilot_executor to strip suffixes using
thinking.ParseSuffix, matching the pattern used by other executors.

Also add test coverage for:
- OAuthModelAliasChannel github-copilot and kiro channel resolution
- Suffix preservation in alias resolution for github-copilot
- normalizeModel suffix stripping in github_copilot_executor
2026-02-10 23:34:19 +03:00
Skyuno
7b01ca0e2e fix(kiro): implement web search MCP integration for streaming and non-streaming paths
Add complete web search functionality that routes pure web_search requests to the Kiro MCP endpoint instead of the normal GAR API.

Executor changes (kiro_executor.go):

- Add web_search detection in Execute() and ExecuteStream() entry points using HasWebSearchTool() to intercept pure web_search requests before normal processing

- Add 'kiro' format passthrough in buildKiroPayloadForFormat() for pre-built payloads used by callKiroRawAndBuffer()

- Implement handleWebSearchStream(): streaming search loop with MCP search -> InjectToolResultsClaude -> callKiroAndBuffer, supporting up to 5 search iterations with model-driven re-search

- Implement handleWebSearch(): non-streaming path that performs single MCP search, injects tool results, calls normal Execute path, and appends server_tool_use indicators to response

- Add helper methods: callKiroAndBuffer(), callKiroRawAndBuffer(), callKiroDirectStream(), sendFallbackText(), executeNonStreamFallback()

Web search core logic (kiro_websearch.go) [NEW]:

- Define MCP JSON-RPC 2.0 types (McpRequest, McpResponse, McpResult, McpContent, McpError)

- Define WebSearchResults/WebSearchResult structs for parsing MCP search results

- HasWebSearchTool(): detect pure web_search requests (single-tool array only)

- ContainsWebSearchTool(): detect web_search in mixed-tool arrays

- ExtractSearchQuery(): parse search query from Claude Code's tool_use message format

- CreateMcpRequest(): build MCP tools/call request with Kiro-compatible ID format

- InjectToolResultsClaude(): append assistant tool_use + user tool_result messages to Claude-format payload for GAR translation pipeline

- InjectToolResults(): modify Kiro-format payload directly with toolResults in currentMessage context

- InjectSearchIndicatorsInResponse(): prepend server_tool_use + web_search_tool_result content blocks to non-streaming response for Claude Code search count display

- ReplaceWebSearchToolDescription(): swap restrictive Kiro tool description with minimal re-search-friendly version

- StripWebSearchTool(): remove web_search from tools array

- FormatSearchContextPrompt() / FormatToolResultText(): format search results for injection

- SSE event generation: SseEvent type, GenerateWebSearchEvents() (11-event sequence), GenerateSearchIndicatorEvents() (server_tool_use + web_search_tool_result pairs)

- Stream analysis: AnalyzeBufferedStream() to detect stop_reason and web_search tool_use in buffered chunks, FilterChunksForClient() to strip tool_use blocks and adjust indices, AdjustSSEChunk() / AdjustStreamIndices() for content block index offset management

MCP API handler (kiro_websearch_handler.go) [NEW]:

- WebSearchHandler struct with MCP endpoint, HTTP client, auth token, fingerprint, and custom auth attributes

- FetchToolDescription(): sync.Once-guarded MCP tools/list call to cache web_search tool description

- GetWebSearchDescription(): thread-safe cached description retrieval

- CallMcpAPI(): MCP API caller with retry logic (exponential backoff, retryable on 502/503/504), AWS-aligned headers via setMcpHeaders()

- ParseSearchResults(): extract WebSearchResults from MCP JSON-RPC response

- setMcpHeaders(): set Content-Type, Kiro agent headers, dynamic fingerprint User-Agent, AWS SDK identifiers, Bearer auth, and custom auth attributes

Claude request translation (kiro_claude_request.go):

- Rename web_search -> remote_web_search in convertClaudeToolsToKiro() with dynamic description from GetWebSearchDescription() or hardcoded fallback

- Rename web_search -> remote_web_search in BuildAssistantMessageStruct() for tool_use content blocks

- Add remoteWebSearchDescription constant as fallback when MCP tools/list hasn't been fetched
2026-02-11 00:02:30 +08:00
Luis Pater
45b6fffd7f Merge branch 'main' into plus 2026-02-09 20:00:16 +08:00
hkfires
49c1740b47 feat(executor): add session ID and HMAC-SHA256 signature generation for iFlow API requests 2026-02-09 19:29:42 +08:00
rico
d468eec6ec fix(copilot): prevent premium request count inflation for Claude models
> Copilot Premium usage significantly amplified when using amp

- Add X-Initiator header (user/agent) based on last message role to
  prevent Copilot from billing all requests as premium user-initiated
- Add flattenAssistantContent() to convert assistant content from array
  to string, preventing Claude from re-answering all previous prompts
- Align Copilot headers (User-Agent, Editor-Version, Openai-Intent) with
  pi-ai reference implementation

Closes #113

Amp-Thread-ID: https://ampcode.com/threads/T-019c392b-736e-7489-a06b-f94f7c75f7c0
Co-authored-by: Amp <amp@ampcode.com>
2026-02-08 02:22:10 +08:00
Luis Pater
eddf31e55b Merge branch 'main' into plus 2026-02-07 09:36:52 +08:00
Luis Pater
7e9d0db6aa Merge pull request #1467 from dusty-du/fix/kimi-toolcall-reasoning-content
Fix Kimi tool-call payload normalization for reasoning_content
2026-02-07 09:35:04 +08:00
Luis Pater
6b83585b53 Merge branch 'router-for-me:main' into main 2026-02-07 08:52:51 +08:00
Luis Pater
78ef04fcf1 fix(kimi): reduce redundant payload cloning and simplify translation calls 2026-02-07 08:51:48 +08:00
Luis Pater
c20507c15e Merge branch 'router-for-me:main' into main 2026-02-07 06:43:17 +08:00
Luis Pater
f7d0019df7 fix(kimi): update base URL and integrate ClaudeExecutor fallback
- Updated `KimiAPIBaseURL` to remove versioning from the root path.
- Integrated `ClaudeExecutor` fallback in `KimiExecutor` methods for compatibility with Claude requests.
- Simplified token counting by delegating to `ClaudeExecutor`.
2026-02-07 06:42:08 +08:00
test
52364af5bf Fix Kimi tool-call reasoning_content normalization 2026-02-06 14:46:16 -05:00
Luis Pater
2e8d49a641 Merge pull request #191 from CheesesNguyen/feat/kiro-api-models-and-context-usage
feat(kiro): add contextUsageEvent handler
2026-02-07 01:33:49 +08:00
Luis Pater
40ad4a42ea Merge branch 'main' into plus 2026-02-06 21:04:32 +08:00
Luis Pater
68cb81a258 feat: add Kimi authentication support and streamline device ID handling
- Introduced `RequestKimiToken` API for Kimi authentication flow.
- Integrated device ID management throughout Kimi-related components.
- Enhanced header management for Kimi API requests with device ID context.
2026-02-06 20:43:30 +08:00
CheesesNguyen
16693053f5 feat(kiro): add contextUsageEvent handler and simplify model structs
- Add contextUsageEvent case handler in kiro_executor.go for both
  parseEventStream and streamToChannel functions
- Handle nested format: {"contextUsageEvent": {"contextUsagePercentage": 0.53}}
- Keep KiroModel struct minimal with only essential fields
- Remove unused KiroPromptCachingInfo struct from kiro_model_converter.go
- Remove unused SupportedInputTypes and PromptCaching fields from KiroAPIModel
2026-02-06 11:12:27 +07:00
test
f5f26f0cbe Add Kimi (Moonshot AI) provider support
- OAuth2 device authorization grant flow (RFC 8628) for authentication
- Streaming and non-streaming chat completions via OpenAI-compatible API
- Models: kimi-k2, kimi-k2-thinking, kimi-k2.5
- CLI `--kimi-login` command for device flow auth
- Token management with automatic refresh
- Thinking/reasoning effort support for thinking-enabled models

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
2026-02-05 19:24:46 -05:00
Luis Pater
165e03f3a7 Merge branch 'router-for-me:main' into main 2026-02-06 05:32:10 +08:00
Luis Pater
b4e034be1c refactor(executor): centralize Codex client version and user agent constants
- Introduced `codexClientVersion` and `codexUserAgent` constants for better maintainability.
- Updated `EnsureHeader` calls to use the new constants.
2026-02-06 05:30:28 +08:00
Joao
84fcebf538 feat: add Claude Opus 4.6 support for Kiro
- Add kiro-claude-opus-4-6 and kiro-claude-opus-4-6-agentic to model registry
- Add model ID mappings for claude-opus-4.6 variants
- Support both kiro- prefix and native format (claude-opus-4.6)
- Tested and working with Kiro API
2026-02-05 21:26:29 +00:00
Luis Pater
74d9a1ffed Merge branch 'router-for-me:main' into main 2026-02-06 03:27:27 +08:00
Luis Pater
a5a25dec57 refactor(translator, executor): remove redundant bytes.Clone calls for improved performance
- Replaced all instances of `bytes.Clone` with direct references to enhance efficiency.
- Simplified payload handling across executors and translators by eliminating unnecessary data duplication.
2026-02-06 03:26:29 +08:00
Luis Pater
e93eebc2e9 Merge branch 'router-for-me:main' into main 2026-02-06 01:53:55 +08:00
Luis Pater
09ecfbcaed refactor(executor): optimize payload cloning and streamline SDK translator usage
- Replaced unnecessary `bytes.Clone` calls for `opts.OriginalRequest` throughout executors.
- Introduced intermediate variable `originalPayloadSource` to simplify payload processing.
- Ensured better clarity and structure in request translation logic.
2026-02-06 01:44:20 +08:00
Luis Pater
6b6b343922 Merge branch 'main' into plus 2026-02-05 19:51:56 +08:00
Luis Pater
25c6b479c7 refactor(util, executor): optimize payload handling and schema processing
- Replaced repetitive string operations with a centralized `escapeGJSONPathKey` function.
- Streamlined handling of JSON schema cleaning for Gemini and Antigravity requests.
- Improved payload management by transitioning from byte slices to strings for processing.
- Removed unnecessary cloning of byte slices in several places.
2026-02-05 19:00:30 +08:00
Skyuno
fcdd91895e Merge remote-tracking branch 'upstream/main' into fix/json-truncation-rework 2026-02-03 20:28:32 +08:00
Luis Pater
5e23975d6e Merge branch 'router-for-me:main' into main 2026-02-03 01:50:45 +08:00
Luis Pater
250f212fa3 fix(executor): handle "global" location in AI platform URL generation 2026-02-03 01:39:57 +08:00
Skyuno
ba168ec003 fix(kiro): skip _partial field (may contain hallucinated paths), add pwd hint for retry 2026-02-02 05:17:39 +08:00
Skyuno
a12e22c66f Revert "Merge pull request #150 from PancakeZik/fix/write-tool-truncation-handling"
This reverts commit fd5b669c87, reversing
changes made to 30d832c9b1.
2026-02-02 05:17:39 +08:00
Luis Pater
38f7e754ca Merge branch 'router-for-me:main' into main 2026-02-01 20:22:46 +08:00
hkfires
ac802a4646 refactor(codex): remove codex instructions injection support 2026-02-01 14:33:31 +08:00
Luis Pater
b43610159f Merge branch 'router-for-me:main' into main 2026-02-01 05:30:36 +08:00
Luis Pater
6d8609e457 feat(config): add payload filter rules to remove JSON paths
Introduce `Filter` rules in the payload configuration to remove specified JSON paths from the payload. Update related helper functions and add examples to `config.example.yaml`.
2026-02-01 05:29:41 +08:00
Luis Pater
dcd0ae7467 Merge branch 'router-for-me:main' into main 2026-01-31 23:49:45 +08:00
Luis Pater
d216adeffc Fixed: #1372 #1366
fix(caching): ensure unique cache_control injection using count validation
2026-01-31 23:48:50 +08:00
Luis Pater
0606a7762c Merge branch 'router-for-me:main' into main 2026-01-31 03:14:11 +08:00
Luis Pater
f887f9985d Merge pull request #1248 from shekohex/feat/responses-compact
feat(openai): add responses/compact support
2026-01-31 03:12:55 +08:00
Luis Pater
e662c020a9 Merge branch 'router-for-me:main' into main 2026-01-31 01:43:48 +08:00
Luis Pater
7ff3936efe fix(caching): ensure prompt-caching beta is always appended and add multi-turn cache control tests 2026-01-31 01:42:58 +08:00
Luis Pater
b1204b1423 Merge branch 'router-for-me:main' into main 2026-01-31 01:15:14 +08:00
Luis Pater
24cf7fa6a2 Merge pull request #156 from taetaetae/fix/kiro-api-region
fix(kiro): Do not use OIDC region for API endpoint
2026-01-31 01:13:47 +08:00
taetaetae
fafef32b9e fix(kiro): Do not use OIDC region for API endpoint
Kiro API endpoints only exist in us-east-1, but OIDC region can vary
by Enterprise user location (e.g., ap-northeast-2 for Korean users).

Previously, when ProfileARN was not available, the code fell back to
using OIDC region for API calls, causing DNS resolution failures:

  lookup codewhisperer.ap-northeast-2.amazonaws.com: no such host

This fix removes the OIDC region fallback for API endpoints.
The region priority is now:
1. api_region (explicit override)
2. ProfileARN region
3. us-east-1 (default)

Fixes: Issue #253 (200-400x slower response times due to DNS failures)
2026-01-31 00:05:53 +09:00
Joao
1e764de0a8 feat(kiro): switch to Amazon Q endpoint as primary
Switch from CodeWhisperer endpoint to Amazon Q endpoint for all auth types:

- Use q.{region}.amazonaws.com/generateAssistantResponse as primary endpoint
- Works universally across all AWS regions (CodeWhisperer only exists in us-east-1)
- Use application/json Content-Type instead of application/x-amz-json-1.0
- Remove X-Amz-Target header for Q endpoint (not required)
- Add x-amzn-kiro-agent-mode: vibe header
- Add x-amzn-codewhisperer-optout: true header
- Keep CodeWhisperer endpoint as fallback for compatibility

This change aligns with Amazon's consolidation of services under the Q branding
and provides better multi-region support for Enterprise/IDC users.
2026-01-30 13:50:19 +00:00
Luis Pater
fcefa2c820 Merge pull request #152 from taetaetae/feat/kiro-dynamic-region-support
feat(kiro): Add dynamic region support for API endpoints
2026-01-30 21:30:04 +08:00
Luis Pater
fd5b669c87 Merge pull request #150 from PancakeZik/fix/write-tool-truncation-handling
fix: handle Write tool truncation when content exceeds API limits
2026-01-30 21:15:31 +08:00
Luis Pater
30d832c9b1 Merge pull request #144 from woopencri/main
fix: handle zero output_tokens for kiro non-streaming requests
2026-01-30 21:06:20 +08:00
Luis Pater
2448691136 Merge pull request #143 from CheesesNguyen/fix/kiro-refresh-token
fix: refresh token for kiro enterprise account
2026-01-30 21:05:00 +08:00