docmancer 0.6.5

Compress local documentation context for coding agents.

Your agents' memory, unified, local, and yours.

Install | First run | What you get | Wiki

---

Your coding agents (Claude Code, Codex, Cursor, Gemini, OpenCode, Cline, Windsurf, and more) already write memory, instructions, and rules all over this machine, each locked inside its own tool. Docmancer discovers all of it, syncs it into one local hybrid (lexical + dense) index, and lets you recall any past decision instantly and offline. The full memory loop is four steps:

1. Sync (docmancer memory sync): discover and index every agent's memory, instructions, and rules into one local SQLite index. Local, offline, no keys.
2. Recall (docmancer memory query): hybrid search across everything your agents have ever written, with source provenance.
3. Consolidate (docmancer memory consolidate): use Mistral AI to turn the scattered, duplicated memory into one coherent, review-only master-memory draft.
4. Apply (docmancer memory apply): materialize the reviewed draft into an agent's always-loaded file, so the context loads every session with no tool call.

The same engine also does docs RAG as a secondary capability: point it at a folder of Markdown / PDF / DOCX / RTF / HTML or a docs URL (GitBook, Mintlify, generic web, GitHub) and query it the same way. A fresh install ships everything you need for the local path: SQLite FTS5 for lexical search, a static embedding model (potion-base-8M) vendored in the package so there is no large model download and no network at runtime, and sqlite-vec for dense vectors in a single local file with no daemon.

Install

pipx install docmancer    # Python 3.11, 3.12, or 3.13

If pipx picks an unsupported interpreter, pin one: pipx install docmancer --python python3.13.

First run

Two commands take you from a fresh install to recalling your agents' memory:

docmancer setup                                  # discovers and indexes the agent memory already on this machine
docmancer memory query "why did we pick Railway" # recall a past decision, offline

setup creates ~/.docmancer/ with the config and SQLite database, syncs the memory, instructions, and rules your coding agents already wrote (Claude Code, Codex, Cursor, Gemini, OpenCode, Cline, Windsurf, and more) plus repo-level CLAUDE.md / AGENTS.md / GEMINI.md, auto-detects installed agents, and installs their skill files. There is no large model download and no network at runtime: the static embedding model is vendored in the package.

Re-sync any time and see exactly what was indexed and from where:

docmancer memory sync                # discover, redact, and (re)index everything
docmancer memory sources             # provenance: agent, type, scope, title, path, char count
docmancer memory sources --preview   # live re-harvest (what WOULD index) without writing

Want docs RAG too? The same engine indexes documentation:

docmancer ingest ./docs                             # index local files
docmancer add https://docs.pytest.org               # or a docs URL
docmancer query "How do I parametrize a fixture?"   # hybrid search across the docs index

Consolidate and carry memory across agents (Mistral AI)

Syncing gives you one searchable index. Consolidation turns that pile into a single coherent memory. docmancer memory consolidate sends your retrieved local memory (privacy-redacted first) to Mistral AI by default and gets back a review-only master-memory draft: deduplicated, grouped into compact sections, with conflicts surfaced as warnings instead of silently resolved.

export MISTRAL_API_KEY=...    # the only extra step; the local commands never need a key
docmancer memory consolidate \
  --query "deployment and infra decisions" \
  --output master-memory-draft.md \
  --draft-quality fast \
  --timeout 180

Once you have reviewed the draft, materialize it into an agent's always-loaded file so the context loads every session with no tool call and, crucially, so memory written in one agent shows up in the others:

docmancer memory apply --agent codex   # uses master-memory-draft.md by default
docmancer memory apply --agent codex --dry-run   # preview the diff first

apply is local and keyless. It writes only inside a clearly delimited managed block, takes a timestamped backup first, and never touches your own surrounding content. --remove strips the block for a clean uninstall. This is the only command that writes consolidated memory into agent-owned files, and it is never automatic. (docmancer install codex / claude-code also inject a short recall instruction into the same files, in their own managed block.)

Mistral is used directly through the official mistralai client: Mistral structured outputs extract durable memory facts, and a Mistral chat model (mistral-small-2506 by default) produces the review-only consolidated draft. Pick any Mistral model your account provisions with --model, or set DOCMANCER_MISTRAL_MODEL to change the default once. Consolidation uses smaller bounded batches by default, --max-output-tokens caps generated output per request, and --draft-quality fast uses more aggressive compression.

OpenRouter is available as an explicit fallback for consolidation when you want another hosted model:

export OPENROUTER_API_KEY=...
docmancer memory consolidate \
  --provider openrouter \
  --model openai/gpt-4.1-nano \
  --output master-memory-draft.md \
  --yes

With OpenRouter, --model accepts any OpenRouter chat model id your account can use, and DOCMANCER_OPENROUTER_MODEL changes the default. Use --timeout, DOCMANCER_MISTRAL_TIMEOUT_SECONDS, or DOCMANCER_OPENROUTER_TIMEOUT_SECONDS to bound each provider request, with a finite 180 second default and 0 for the provider default. Optionally, mistral-embed-2312 can build the local vector index (docmancer init --embedding-provider mistral). Every cloud-backed command fails gracefully with a clear message when the provider key is not set or the API call fails, prints a cloud-use notice before the first call, sends a tiny preflight chat request before large memory payloads, logs each request before sending it, and runs secret redaction before any text leaves your machine. See the Configuration and Commands pages for details.

What you get

Your agents' memory, unified. docmancer memory sync discovers and indexes the memory, instructions, and rules your coding agents already wrote (Claude Code, Codex, Cursor, Gemini, OpenCode, Cline, Windsurf, and more, plus repo-level CLAUDE.md / AGENTS.md / GEMINI.md), then answers questions about them through one local index. docmancer memory sources shows exact provenance per file. The local path uploads nothing.

Consolidate with Mistral AI. docmancer memory consolidate turns the scattered index into one review-only master-memory draft via direct Mistral by default, and docmancer memory apply bakes the reviewed result into an agent's always-loaded file so context carries across agents. OpenRouter is available as an explicit fallback with --provider openrouter --model <model-id>. Key-gated, privacy-redacted, and review-only.

Callable over MCP. The packaged docmancer-mcp stdio server exposes local memory and docs search to MCP clients. docmancer mcp install codex (or claude-code, claude-desktop) wires it up; optional Mistral tools appear when MISTRAL_API_KEY is set. Requires the mcp extra.

Hybrid search by default. query and memory query fan out across SQLite FTS5 (lexical, BM25-reranked) and dense vectors from a vendored static model (potion-base-8M) in sqlite-vec, then fuse results with Reciprocal Rank Fusion. Sparse (SPLADE) signals are available on the optional heavy Qdrant backend. The token budget keeps responses small so your agent has room for actual work:

Context pack: ~900 tokens vs ~4800 raw docs tokens (81.2% less docs overhead, 5.33x agentic runway)

No large model download, offline at runtime. The static embedding model ships inside the wheel, so there are no API keys and no network needed to embed or query. Optional OpenAI / Voyage / Cohere providers exist if you want them; a heavier FastEmbed + Qdrant backend is available via pipx install "docmancer[embeddings-heavy]".

Where your data lives and how to remove it

The local memory index is stored in SQLite-backed files under ~/.docmancer/ (override the main database with DOCMANCER_MEMORY_DB). Sync, query, status, sources, apply, and clear run locally. Mistral-backed commands are optional, key-gated, and send selected memory text only after privacy redaction and a cloud-use confirmation. You can preview exactly what would be indexed with docmancer memory sync --dry-run, scope the harvest with --include / --exclude globs, and delete the local memory index files with docmancer memory clear. There is no telemetry and no phone-home.

Inspectable. Every section is written to ~/.docmancer/extracted/ as Markdown plus JSON. docmancer inspect shows index stats. docmancer query --explain shows which signal (lexical / dense / sparse) placed each result.

Agent integration built in. docmancer setup drops skill files for Claude Code, Cursor, Codex, Cline, Claude Desktop, Gemini, GitHub Copilot, and OpenCode. For Claude Code and Codex it also injects a memory-recall instruction into the always-loaded CLAUDE.md / ~/.codex/AGENTS.md (a managed block), so the agent reliably calls docmancer memory query before answering questions about past work.

Where to next

The wiki is the authoritative reference for everything else. Pick a page based on what you need:

Page	When to read it
Commands	Core docs commands and Qdrant lifecycle commands
Configuration	All YAML keys, env vars, and the API-key reference
Architecture	How ingest, retrieval, and Qdrant lifecycle work
Supported Sources	What file formats and URL providers are covered
Install Targets	Where each agent's skill file lands
Troubleshooting	Common errors and fixes

Wiki home | Changelog | PyPI