[pull] main from Kilo-Org:main #366
Merged
Conversation
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
- Add supportsFim(), completeFim(), and streamFim() methods to MistralHandler - FIM support is only enabled for codestral models (codestral-latest, codestral-2405, etc.) - Uses Mistral's v1/fim/completions endpoint with proper base URL routing - Supports both streaming and non-streaming FIM completions - Add comprehensive test coverage with 10 test cases - Update continuedev Mistral LLM to accurately check model support - Implementation follows the same pattern as KiloCode OpenRouter handler Closes: FIM functionality now works for Mistral's codestral models
- Add zai-glm-4.7 to Cerebras models in @roo-code/types\n- Update Cerebras docs to include GLM 4.7 and note GLM 4.6 deprecation\n- Add changeset for release notes
- Add FimCapableHandler interface for duck typing - Add isFimCapable() type guard function - Update supportsFim() and generateFimResponse() to use duck typing - Any handler implementing supportsFim() and streamFim() now works with FIM - Enables MistralHandler FIM support without explicit class checks
Address review comment about signature mismatch with IFimProvider interface. The streamFim method now accepts optional taskId and onUsage parameters, and calls onUsage when usage data is available in the stream response.
Add detection for suggestions where the tail is repeating itself - a common LLM failure mode. For example, when a model gets stuck in a loop and suggests 'the beginning. We are going to start from the beginning. We are going to start from the beginning...' repeated many times. The new containsRepetitivePhraseFromPrefix function looks for phrases (20-80 chars) from the end of the suggestion that appear 3+ times in the suggestion text, indicating the model got stuck in a repetition loop.
…suggestions feat(ghost): filter repetitive phrase suggestions in autocomplete
- Remove completeFim from IFimProvider interface - Remove completeFim implementation from MistralHandler - Remove completeFim implementation from KilocodeOpenrouterHandler - Remove completeFim tests from both test files The method was only used in tests and not in production code. streamFim is the only FIM method needed.
add new file marker to hopefully prevent further markers
feat(mistral): add FIM support for codestral models
…red-filter refactor: use shared postprocessGhostSuggestion filter in ChatTextAreaAutocomplete
…ema doesn't yet support it Discussion: https://kilo-code.slack.com/archives/C08P0HYC9S4/p1767727278860319
Disable structured outputs for Anthropic models, because the tool schema doesn't yet support it
Changeset version bump
Improved prompt caching when using Anthropic models on OpenRouter with native tool calling
…odel docs(types/cerebras): add zai-glm-4.7
Sign up for free
to subscribe to this conversation on GitHub.
Already have an account?
Sign in.
3 participants
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.
See Commits and Changes for more details.
Created by
pull[bot] (v2.0.0-alpha.4)
Can you help keep this open source service alive? 💖 Please sponsor : )