actx ingest
actx ingest
Section titled “actx ingest”Ingest files and directories into the knowledge plane, making them available for semantic search via ?k queries.
actx ingest <source> [options]Arguments
Section titled “Arguments”| Argument | Required | Description |
|---|---|---|
source | ✅ | File or directory path to ingest |
What It Does
Section titled “What It Does”- Scans the source path for supported file types
- Chunks each file into semantically meaningful segments
- Passes each chunk through the knowledge write gate (deduplication + validation)
- Stores chunks in the knowledge plane (SurrealDB with HNSW vector indices)
Supported Formats
Section titled “Supported Formats”| Format | Extensions |
|---|---|
| Markdown | .md, .mdx |
| TypeScript/JavaScript | .ts, .js, .tsx, .jsx |
| Python | .py |
| Rust | .rs |
| YAML/JSON | .yaml, .yml, .json |
| Plain text | .txt |
Examples
Section titled “Examples”Ingest a directory
Section titled “Ingest a directory”actx ingest docs/Ingest a single file
Section titled “Ingest a single file”actx ingest README.mdThen search
Section titled “Then search”actx query '?k "authentication" #code ^3'- The knowledge filesystem watcher (configured in
actx.yaml) automatically re-indexes when source files change - Duplicate content is deduplicated via content-addressed hashing
- Large files are chunked to optimize for semantic search relevance
See Also
Section titled “See Also”- actx query — search the ingested knowledge
- Seven Planes — the knowledge plane architecture