knowledge-index 0.5.0a0

Search index for agent memory — knowledge graph index for your documents.

Knowledge Index (ki)

The safe, easy, fast search engine for your knowledge base.

Point ki at a folder of markdown — notes, docs, a wiki — and it syncs to a Neo4j knowledge-graph index you can search and navigate in seconds, from the CLI or any AI agent. Agents and humans alike go understand → search → retrieve with no prior expertise.

ki is a search index, not a store. It only reflects what's on your filesystem — source files are never modified (safe on an Obsidian vault, a git repo, a research folder), the index is disposable cache you can rebuild with one ki index, and there's no LLM work or embeddings at index time (vendor-neutral, instant to set up, free). See docs/general-philosophy.md for the tenets.

Install

curl -sSfL https://knowledge-index.ai/install.sh | bash   # ki + neo4j-cli + agent skills
ki configure                                              # one-time Neo4j: Local (Podman), Aura, or Existing

Prefer Python tooling? uv tool install knowledge-index (or uvx knowledge-index --help to try without installing).

Quickstart

cd ~/my-notes
ki index . --profile personal     # sync the folder into the graph (first index binds a profile)
ki outline my-notes --full        # table-of-contents view of the vault
ki search "rate limiting"         # find the right slice
ki get --type full "<uri>"        # read it (copy a uri from outline/search)

ki writes one marker per vault — .ki/vault.yaml, holding the vault's slug uri, its bound profile name, and an optional description. It's safe to commit (no credentials). Re-running ki index resyncs; nuke the graph and rebuild anytime.

With an AI agent

Coding agents (Claude Code, Cursor, Windsurf, …) shell out to ki directly. ki skill install drops the bundled knowledge-base skill into each detected agent's config so it knows when to reach for ki — "what did I write about X?", "build a knowledge base from this folder," "find related material" — and when not to.

ki skill install      # into all detected agents (the installer already does this)
ki skill list         # what's wired up / detected

The shipped skill scopes to the local, single-vault workflow; see docs/skills.md.

Usage patterns

ki is optimized for one local vault at a time and generalizes from there. The full model — profiles, vaults, and how every command resolves scope — is docs/scoping.md.

1. Local, single vault (the default, ~95%) — your markdown on disk; the full lifecycle (index/update + search/retrieve). No flags: ki uses the vault you're standing in.
2. Remote vaults — read-only search/retrieval over a KB you don't have locally (sharing a KB, powering apps/production). Name it explicitly with --profile (+ --vault).
3. Cross-vault search — query several vaults in one profile at once (--profile; navigation/read commands only).

Commands

ki <cmd> --help is the source of truth for flags.

• Index / update: index (full sync) — add / rm / mv (incremental) are planned.
• Read / retrieve: search, outline (alias tree), get, status. Depth docs: search · get · outline.
• Remove from index: drop (one vault) · nuke (a whole profile). Source files are never touched.
• Admin / info: configure, profile list, vault list, skill.

ki search is one ranked fulltext sweep over documents + sections (no embeddings) — narrow with --types, scope locally with --under or remotely with --profile / --vault. See docs/commands/search.md.

Neo4j connection

ki configure offers three paths: Local (Podman neo4j:latest + APOC/GenAI — full runbook in skills/knowledge-base/references/neo4j-podman.md), Aura (billable cloud via neo4j-cli; never picked without consent), and Existing (any Bolt URI). Credentials live in ~/.config/ki/config.yaml (mode 0600); vaults reference profiles by name, so syncing a vault folder never leaks them.

Limitations & roadmap

• Markdown only — convert PDF / docx / HTML first (pandoc, markitdown), then ki index.
• No vector search yet — fulltext is the substrate (by design — see the philosophy); hybrid is roadmap.
• No MCP server yet — coding agents work today; chat apps (claude.ai, ChatGPT, …) need an MCP bridge (roadmap).
• Plaintext passwords (config.yaml, mode 0600) — OS keyring is roadmap.

Full roadmap → open issues.

Docs

Start at docs/: architecture · general-philosophy · scoping · data model. Contributors: AGENTS.md.

Development

git clone https://github.com/zach-blumenfeld/knowledge-index.git && cd knowledge-index
uv sync --extra dev
uv run pytest                                   # integration tests auto-skip without a Neo4j
uv run ruff check src/ tests/ scripts/

Point integration tests at any Neo4j via KI_TEST_NEO4J_URI / KI_TEST_NEO4J_USER / KI_TEST_NEO4J_PASSWORD (the suite is destructive — don't use a Neo4j with real data). Design principles, project map, test fixtures, and the release flow live in AGENTS.md.

License

See LICENSE.