with one click
abd-embed-vectors
// Embed text chunks into a local FAISS vector index for semantic search. Use when the user wants to "embed chunks", "build a vector index", "create embeddings", or prepare chunked content for RAG retrieval.
// Embed text chunks into a local FAISS vector index for semantic search. Use when the user wants to "embed chunks", "build a vector index", "create embeddings", or prepare chunked content for RAG retrieval.
[HINT] Download the complete skill directory including SKILL.md and all related files
| name | abd-embed-vectors |
| description | Embed text chunks into a local FAISS vector index for semantic search. Use when the user wants to "embed chunks", "build a vector index", "create embeddings", or prepare chunked content for RAG retrieval. |
| license | MIT |
| metadata | {"author":"agilebydesign","version":"1.0.0"} |
Turn chunked markdown into vectors and store them in a local FAISS index so semantic search can return the right passages for a question — not only keyword matches.
python scripts/embed_and_index.py --path <source_folder>/memory
python scripts/embed_and_index.py --path <source_folder>/memory --replace
The FAISS index is written to <source_folder>/memory/rag/ — next to the chunks, not inside the skill directory.
<source>/memory/rag/
index.faiss # vector index
metadata.json # chunk paths and metadata
Full detail: references/config.md. Agent workspace policy: AGENTS.md (Workspace (topic root) — config first).
Set OPENAI_API_KEY for the embedding model. Prefer agents/abd-context-to-memory/conf/.secrets (copy from conf/.secrets.example). The config module loads conf/.secrets → conf/.env → skill/orchestrator .env files → cwd/.env automatically.
Set CONTENT_MEMORY_ROOT= in conf/.secrets (recommended), pass --path, or cd to the corpus folder. When --path is omitted, embed uses <ROOT>/memory by default.
pip install -r requirements-rag.txt
Requires openai, faiss-cpu, and related packages.
<source>/memory/rag/), not inside the skill package.--replace to avoid stale vectors.memory/ folder; use chunk_inputs in the chunking spec (see abd-chunk-markdown) to control what gets chunked before embedding.