Rewrote embed-builtin-knowledge.py to dynamically discover all project
docs, internal docs, identity files, and skill definitions instead of
using a hardcoded 6-chunk array. Artifact now covers full corpus via
OpenRouter BAAI/bge-m3 at 1024 dimensions.
Also fixed stale config.ts defaults: EMBED_MODEL from nomic-embed to
BAAI/bge-m3, EMBED_DIMENSIONS from 768 to 1024 (matching the pgvector
schema's vector(1024) column).
Tested: imported into live clawdie_skills DB, FTS search confirmed.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
Build: pass | Tests: pass — Tests 603 passed (603)
- Add embed-builtin-knowledge.py script for OpenRouter bge-m3 embeddings
- Update artifact.sql with pre-generated 1024-dim embeddings for 6 chunks
- Update metadata: openrouter/baai/bge-m3, embedding_count=6, search_mode=hybrid
- Change locale from en-US to sl-SI per project language setting
Author: Clawdie AI <hello@clawdie.si>
Co-Authored-By: Samo Blatnik, Zed.ai
- Replace hardcoded "klavdija" tmux session names with $AGENT_NAME in
glass.sh and run-klavdija.sh
- Replace hardcoded 192.168.100.x fallbacks with 10.0.1.x throughout
scripts/memory/ and heartbeat.sh defaults
- Fix heartbeat PGPASSWORD: use inline `VAR=val cmd` form so psql
actually receives the password in the subshell
- Fix heartbeat curl hangs: add --max-time to all curl calls
- Add run-klavdija.sh and src/transcription.ts (previously untracked)
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
Build: pass | Tests: pass — Tests 431 passed (431)
- Remove 7 redundant .env vars: DB_NAME, DB_USER, DB_HOST, DB_PORT,
MEMORY_DB_NAME, MEMORY_DB_USER, MEMORY_DB_URL — all now derived from
AGENT_NAME by config.ts and common.sh
- Fix DB_HOST in common.sh pointing to .2 (controlplane) instead of .3 (db)
- common.sh: normalise AGENT_NAME → Postgres identifier, same algorithm
as db-identifiers.ts; embed config now reads from .env instead of
overriding with stale OpenRouter values
- embed.sh: drop OPENROUTER_API_KEY requirement; use EMBED_BASE_URL +
EMBED_API_KEY (empty = local llama-server, no auth needed)
- memory-hydrate-pg.sh, memory-lifecycle.ts: replace ai_brain/clawdie_brain
literals with live DB_NAME / MEMORY_DB_NAME values
Bump to 0.9.1.
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
---
Build: pass | Tests: pass — Tests 431 passed (431)
Three scripts in scripts/memory/:
- common.sh: shared config, DB connection, dependency checks
- embed.sh: generate 1536d embeddings via OpenRouter API
- chunk.sh: split text on sentence boundaries (~500 chars)
- memory-pg.sh: full CLI for store/search/recent/important/count
Store pipeline: text → chunk → embed → insert (fully automated).
Search: hybrid RRF combining full-text and vector similarity.
All scripts use #!/usr/bin/env bash for FreeBSD compatibility.
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
