v0.2.0

What's Changed

• fix(mcp): drop unexpandable MEMORY_DATA_DIR from Claude Code template by @meshin-dev in #1
• fix(wiki-store): stop YAML line-folding so long scalars stay single-line by @meshin-dev in #3
• fix(flush): make memory flush deterministic via a detached worker by @meshin-dev in #4
• feat(discipline): strengthen always-on routing (attempt-first, wiki-is-default) by @meshin-dev in #6
• chore(deps): bump @ctxr/skill-llm-wiki to ^1.2.0 by @meshin-dev in #8
• feat(wiki-store): nest categories by metadata facets (no flat piles) by @meshin-dev in #9
• chore(deps): bump @ctxr/skill-llm-wiki to ^1.3.0 by @meshin-dev in #10
• feat(memory): fix default recall + navigability (project_module/area split) by @meshin-dev in #11
• chore(deps): bump @ctxr/skill-llm-wiki to ^1.3.1 by @meshin-dev in #12
• feat(embed): default to bge-large-en-v1.5 + model-selection table by @meshin-dev in #13
• feat(mcp): self-refreshing server + facet inference (never unknown area/atom_type) by @meshin-dev in #14
• Feat/yaml driven layout and path override by @meshin-dev in #15
• fix(consolidate): never relocate a leaf when stamping bookkeeping (DUP-ID) by @meshin-dev in #16
• fix(health): report the resolved chain head, not a .env-configured provider by @meshin-dev in #17

New Contributors

• @meshin-dev made their first contribution in #1

Full Changelog: https://github.com/ctxr-dev/llm-wiki-memory/commits/v0.2.0