lorekeep 0.1.7

Temporal knowledge graph for AI agents via MCP

Lorekeep

A temporal knowledge graph for AI agents, over MCP — agents read at query time, contribute at compile time (runtime write planned for phase 2).

Lorekeep compiles a team's raw documentation into a versioned, time-aware knowledge graph (facts.jsonl) and exposes it to coding agents (Claude Code, Cursor, Codex) through the Model Context Protocol — with per-namespace permission and zero servers to run.

It applies Andrej Karpathy's "LLM Knowledge Base" idea: raw docs are the source code, the compiled graph is the executable. Knowledge is processed once at compile time, not re-RAG'd per query — and agent conversations continuously enrich the graph through append-only journals.

---

Why

Existing tools each miss part of what a team needs:

	file-based	temporal KG	compile step	team permission	MCP
Obsidian + MCP	✅	❌	❌	❌	✅
mcp-knowledge-graph	✅	❌	❌	❌ (local)	✅
mem0 / cognee	❌ (DB)	partial	❌	partial (DB)	✅

Lorekeep targets the gap: strictly file-based + temporal graph + compile-once + namespace-scoped permission + MCP — for team-level (not just single-user) knowledge.

Features

• Append-and-resolve [planned] — three write paths (raw/ compile, agent propose, import sessions) converge into one resolve step. Journals are append-only; resolve is pure logic, zero LLM cost.
• Agent-driven knowledge [planned] — agents propose facts at runtime via MCP write tools at zero marginal LLM cost. Confidence-gated: high-confidence auto-merge, low-confidence quarantine.
• File-sovereign — facts.jsonl (one fact per line, sorted) is the single source of truth and the sync unit (git or S3). No binary store committed.
• Temporal — every fact carries valid_from/valid_to (half-open [from, to)); query "what was true at T", history, diffs.
• Namespace permission — facts are tagged ns from the directory tree (raw/<ns>/); agents scoped to namespaces; cross-namespace edges hidden unless both endpoints are visible. Deny-by-default.
• MCP, stdio-first — lorekeep serve exposes 8 read tools (5 write tools planned); lorekeep mcp add wires Claude Code / Cursor / Codex.
• Autonomous agent [planned] — lorekeep agent watch keeps the graph current: auto-compile on raw/ change, auto-resolve pending journals, nightly lint, weekly suggestions.
• Lazy-reload — graph updates (compile or resolve) are visible on the next query. Connect once, use forever.
• Provider-pluggable extraction — litellm (OpenAI / Anthropic / DashScope/Qwen / Ollama). Strict-privacy → Ollama, fully local.
• Tier-1 eval — extraction P/R/F1 vs a gold corpus, entity-resolution F1, graph-structure metrics, determinism property tests.

Install

# from PyPI:
uvx lorekeep init                 # try it without installing

# or from a clone:
git clone https://github.com/manhhailua/lorekeep && cd lorekeep
uv tool install .                 # installs the `lorekeep` command

Quickstart

# 1. bootstrap a data home (~/.config/lorekeep + ~/.local/share/lorekeep)
uvx lorekeep init

# 2. add docs under the data home's raw/<namespace>/
mkdir -p ~/.local/share/lorekeep/raw/backend
cp your-docs.md ~/.local/share/lorekeep/raw/backend/

# 3. set a provider (edit ~/.config/lorekeep/config.yaml), then compile
uvx lorekeep compile                # raw/*.md -> graph/facts.jsonl

# 4. wire a coding agent (writes a portable .mcp.json)
uvx lorekeep mcp add --agent claude --ns backend

# 5. verify
uvx lorekeep doctor

Restart Claude Code → 8 Lorekeep read tools are available, scoped to your namespace.

How it works

               THREE WRITE PATHS                            SYNC
               ════════════════
raw/<ns>/*.md ──► ingest ──► extract(LLM) ──┐
                                            │
agent propose ──► MCP write tools ──► ──────┤  [planned phase 2]
  (ZERO LLM cost, journal append)           │
                                            ├──► resolve ──► writer ──► facts.jsonl
import ──► raw/ ──► compile ────────────────┘    (pure logic,          │
                                                   ZERO LLM)            │
                                                        ┌───────────────┘
                                                        ▼ (git / S3 sync)
               SERVE + QUERY (runtime, per device)
facts.jsonl ──load──► GraphStore ──► ScopedGraph(ns) ──► MCP ──► agent
                         ▲              ▲                      │
                         │              │         ◄── read queries
                          │              └────────── write proposals (journal) [planned]
                         └── lazy-reload on mtime change

               AUTONOMOUS AGENT (daemon) [planned phase 2]
               lorekeep agent watch:
                 ├── watch raw/ → auto-compile
                 ├── periodic resolve → merge journals
                 ├── nightly lint → health check
                 └── weekly suggest → gaps, improvements

Three write paths → one resolve: markdown is compiled by an LLM (chunked + cached); agents will propose facts at runtime through MCP write tools at zero marginal LLM cost (the agent already ran the LLM for the conversation) — planned for phase 2; agent sessions are imported into raw/. All converge at resolve — pure Python logic that merges, deduplicates, validates, and writes byte-stable facts.jsonl.

Serve: GraphStore loads facts.jsonl into a networkx graph with temporal queries. ScopedGraph is the single permission chokepoint — every query is filtered through strict visibility rules. The FastMCP server exposes 8 read tools (5 write tools planned for phase 2) over ScopedGraph. It lazy-reloads when facts.jsonl changes, so compile is instantly visible without reconnecting.

Concepts

fact — one line of facts.jsonl, a node or edge:

{"kind":"node","id":"svc:payments","type":"service","ns":["backend"],"valid_from":"2024-01-15","valid_to":null,"props":{"lang":"go"},"src":["raw/backend/payments.md:12"]}
{"kind":"edge","id":"e_depends_on_0001","type":"depends_on","from":"svc:payments","to":"svc:auth","ns":["backend"],"valid_from":"2024-01-15","valid_to":"2025-03-01","props":{},"src":["...:20"]}

• ns — namespace set; ["public"] is globally visible.
• valid_to: null ⇒ current. History = multiple edges, same endpoints, different windows.
• src — provenance to raw doc line (audit, incremental re-compile, agent citations).

Permission — effective_ns = allowed ∪ {public}. Node visible iff ns ∩ effective_ns ≠ ∅. Edge visible iff both endpoints visible and edge.ns ∩ effective_ns ≠ ∅. Deny-by-default; an edge never reveals a neighbor the caller can't see.

Temporal queries — at_time(T) (snapshot of facts valid at T, half-open [from,to)), history(id) (versions of an entity), changes(t1,t2) (edges that began/ended in the window).

Agent-driven knowledge [planned] — agents will propose facts at runtime through MCP write tools (zero LLM cost). Facts land in pending/<ns>/journal.jsonl with agent id, confidence score, and timestamp. Resolve merges them into the graph: high-confidence (≥0.8) auto-merge, medium (0.5-0.8) merge + flag, low (<0.5) quarantine.

Autonomous agent [planned] — lorekeep agent watch keeps the graph current: watches raw/ for changes → auto-compile; monitors pending/ → auto-resolve; nightly semantic lint; weekly gap suggestions. See docs/architecture/agent.md.

MCP tools (8 read, scoped; 5 write planned)

Read: search · get_node · neighbors · at_time · history · changes · list_namespaces · schema.

Write (journal-based, zero LLM cost, planned phase 2): propose_fact · link_facts · flag_contradiction · update_fact · suggest_improvement.

Every result is filtered to the caller's namespace. Write tools will append to pending/ journals; facts enter the graph on the next resolve pass.

Configuration

config.yaml (resolved by precedence: explicit LOREKEEP_* env > LOREKEEP_HOME > dev marker > XDG):

provider:
  model: openai/qwen-plus                              # litellm model string
  api_base: https://dashscope-intl.aliyuncs.com/compatible-mode/v1
  api_key_env: DASHSCOPE_API_KEY                       # env var name (preferred)
  api_key: null                                        # or inline (gitignored config only)
ns:
  default: [public]
install_source: pypi                                   # pypi = portable .mcp.json

API keys never live in committed files — use api_key_env (env) or inline api_key in the gitignored config only. Examples (DashScope / OpenAI / Ollama) in .lorekeep/config.yaml.example.

Data home & dev mode

Path resolution (high → low): explicit LOREKEEP_* env → LOREKEEP_HOME → dev mode (.lorekeep/ or raw/ in CWD; auto-detected in a source checkout) → XDG (~/.config/lorekeep, ~/.local/share/lorekeep).

Full details, per-path overrides, and lorekeep init: docs/guides/data-home.md. For usage, see the docs/ index.

Evaluation

Tier-1 (CI): extraction P/R/F1 vs a gold corpus, entity-resolution pairwise F1, graph-structure metrics, determinism. Run: uvx lorekeep eval. The north star is systematic thinking with complete information — memory-recall benchmarks (LoCoMo, LongMemEval) are parity checks, not the optimization target. See docs/architecture/evaluation.md.

Project layout

src/lorekeep/
  models.py            shared contract (Node/Edge/Schema/Manifest)
  facts_io.py          facts.jsonl loader (store + eval)
  paths.py             4-tier path resolution (env/home/dev/XDG)
  defaults.py          default schema + config (for `init`)
  config.py, schema_io.py
  compile/{ingest,extract,resolve,writer}.py    the compile pipeline
  compile/providers.py                          LLMProvider (Fake/LiteLLM)
  journal.py           append-only journal writer + loader [planned phase 2]
  agent.py             autonomous agent CLI + daemon [planned phase 2]
  store/{graph,fts}.py                          GraphStore + optional FTS cache
  perm/ns.py                                    ScopedGraph permission chokepoint
  mcp_server.py                                 FastMCP + 8 read tools (5 write planned)
  integrations/{claude_code,cursor,codex,common}.py
  pipeline.py, cli.py
  eval/{gold,construction,retrieval}.py
tests/                 ~140 tests
docs/                  README.md index, architecture/, guides/

Status

v1 (implemented) — compile pipeline + serve (store/permission/MCP read/integrations) + import + data-home + dev mode + lazy-reload + eval. Published to PyPI as lorekeep.

Phase 2 (planned) — journal (append-only pending) + MCP write tools + agent daemon + wiki.md views (Obsidian-compatible markdown output), streamable-HTTP team server, OIDC/SSO, embeddings/hybrid search, full Tier-2 benchmark datasets (HotpotQA/CronQuestions) and the bespoke Tier-3 Lorekeep-Reason eval.

Documentation

The docs/ index is the entry point.

Guides

• Importing agent sessions
• Compiling the knowledge graph
• Serving the graph to coding agents
• Data home & path resolution

Architecture

• Overview · Data model · Pipeline · Journal · Agent · Permission · Temporal · Serve & MCP · Testing & evaluation

License

Lorekeep is released under the MIT License — see LICENSE.

Copyright © 2026 Manh Pham. You're free to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the software, provided the copyright and permission notice are included in all copies. The software is provided "as is", without warranty of any kind.