Update CHANGELOG.md following TalkPipe conventions. Use after implementing a feature, fixing a bug, or making any user-facing change. Ensures correct "update vs add" logic and entry format.
86
87%
Does it follow best practices?
Impact
70%
0.93xAverage score across 3 eval scenarios
Passed
No known issues
Updates CHANGELOG.md under the ## Unreleased section following project conventions.
Is this a minor change to an existing feature already mentioned in Unreleased?
Is this purely internal/refactor with no user impact?
## Unreleased heading)makevectordatabase, run_doc_examples)[text](path) for doc references when helpfulNew feature (add new bullet):
- Added **mySegment** for filtering items by score threshold.Minor change to existing feature (update existing bullet): If Unreleased already has:
- Doc examples tests clean up artifacts (e.g. `my_knowledge_base` from RAG examples) after each test.And you add validation for path traversal, update it:
- Doc examples tests clean up artifacts (e.g. `my_knowledge_base` from RAG examples) after each test.
Cleanup validates paths before deletion to prevent path traversal.Bug fix:
- Fixed chatterlang_serve stream interface so search results display reliably.Multiple related changes (one bullet):
- RAG source citation: ProcessDocumentsSegment preserves title and source (path) as separate
metadata fields; construct_background prioritizes them in prompts; default system prompt and
prompt_directive instruct the LLM to cite sources.0cf1686
If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.