refactor: modularize fetch wrapper and plugin architecture#36
Merged
refactor: modularize fetch wrapper and plugin architecture#36
Conversation
…ontext pruning - Add position-based tool call correlation for Google/Gemini format (body.contents with functionResponse) - Add OpenAI Responses API support (body.input with function_call_output and call_id) - Extract getAllPrunedIds helper for reuse across all format handlers - Build position mapping in chat.params hook for google/google-vertex providers - Remove incomplete strip-reasoning strategy from default config
- Use string instead of nested object for functionResponse.response - Preserves thoughtSignature via spread operator (required for Gemini 3 Pro) - Aligns with Google's flexible response format
…rate modules - Split monolithic index.ts into focused modules for better maintainability - Extract fetch wrapper into lib/fetch-wrapper/ with format-specific handlers: - openai-chat.ts: OpenAI Chat Completions & Anthropic format - openai-responses.ts: OpenAI Responses API format (GPT-5) - gemini.ts: Google/Gemini format with position-based correlation - types.ts: Shared types and utility functions - Create lib/state.ts for centralized plugin state management - Create lib/hooks.ts for event and chat.params handlers - Create lib/pruning-tool.ts for context_pruning tool definition - Create lib/tool-cache.ts for tool parameter caching utilities - Reduce index.ts from 500+ lines to ~80 lines of initialization code
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
Changes
Architecture Improvements
lib/fetch-wrapper/with format-specific handlers:openai-chat.ts: OpenAI Chat Completions & Anthropic formatopenai-responses.ts: OpenAI Responses API format (GPT-5)gemini.ts: Google/Gemini format with position-based correlationtypes.ts: Shared types and utility functionslib/state.tsfor centralized plugin state managementlib/hooks.tsfor event and chat.params handlerslib/pruning-tool.tsfor context_pruning tool definitionlib/tool-cache.tsfor tool parameter caching utilitiesBug Fix
Version
0.3.21