0.7.1

Bug Fixes 🐛

• Sanitize pending/running tool parts to prevent orphaned tool_use by @BYK in #54

0.7.0

New Features ✨

• Add Voyage AI embedding search by @BYK in #50
• Add search config surface and LLM query expansion by @BYK in #49
• Replace forSession() scoring with FTS5 BM25 by @BYK in #48
• Add RRF score fusion and rewrite recall tool by @BYK in #47
• Improve FTS5 search foundations by @BYK in #46

Bug Fixes 🐛

• Catch unhandled exceptions in transform hooks and avoid loading embedding BLOBs by @BYK in #52
• Auto-enable embeddings when VOYAGE_API_KEY is set by @BYK in #51

0.6.2

Bug Fixes 🐛

• Prevent EROFS crash when launched outside a git repo by @BYK in #43
• Cache LTM per session to preserve prompt cache hit rate by @BYK in #42

0.6.1

Bug Fixes 🐛

• Rotate worker sessions after each LLM call and add recall error handling by @BYK in #38

0.6.0

New Features ✨

• Research-informed compaction improvements by @BYK in #35

Bug Fixes 🐛

• Prevent cross-project knowledge entries from leaking into AGENTS.md by @BYK in #36

0.5.3

Bug Fixes 🐛

• Prevent excessive background LLM requests causing rate limiting and sluggishness by @BYK in #33
• Upgrade zod from v3 to v4 by @BYK in #32
• Drop trailing pure-text assistant messages at layer 0 too by @BYK in #31
• Isolate test suite from live lore DB via LORE_DB_PATH preload by @BYK in b96cb956

Internal Changes 🔧

• Remove leaked test fixture entries from AGENTS.md by @BYK in 54dbae18

0.5.2

Bug Fixes 🐛

• Make AGENTS.md export merge-friendly with sorted entries and blank line separators by @BYK in #28

Other

• release: 0.5.2 by @BYK in #29

0.5.1

• Suppress informational log messages in TUI mode by @BYK in #26

0.5.0

New Features ✨

• Add knowledge.enabled option to disable LTM system by @BYK in #23

0.4.4

Bug Fixes 🐛

• (gradient) Persist forceMinLayer and auto-recover from context overflow by @BYK in #21