mempalace-rs

A local-first memory retrieval engine for coding agents, implemented in Rust.

This project stores verbatim project and conversation memory, embeds it locally, and retrieves source-grounded context through MCP. It is built for coding agents that need to remember decisions, prior fixes, commands, project conventions, and user preferences across sessions without running a separate vector database.

What It Does

• Stores project files and conversation turns in a local SQLite database.
• Generates local embeddings with ONNX Runtime and all-MiniLM-L6-v2.
• Retrieves memories with hybrid semantic/BM25 search plus coding-agent intent boosts.
• Sanitizes agent-generated query dumps before retrieval.
• Returns source-grounded results with score provenance and nearby source context.
• Provides a knowledge graph for temporal entity relationships.
• Exposes MCP tools for assistants that support Model Context Protocol.
• Offers a small Rust library API for embedding memory into other services.

Storage

Collapses Python's dual-store (ChromaDB + SQLite) into one file at ~/.mempalace/palace.db:

Table	Purpose
drawers	Text content + embedding BLOB + metadata
entities	KG entity nodes
triples	KG temporal relationship edges

Embeddings are stored as f32 vectors from all-MiniLM-L6-v2. Search uses local cosine similarity over the stored vectors.

---

Benchmarks

Coding-Agent Memory Eval

The repository includes a focused eval fixture for practical coding-agent memory questions. It stores realistic memories about project decisions, prior failures, commands, conventions, user preferences, and current direction, then asks 40 questions such as:

• why did we choose bundled sqlite?
• how did we fix the migration test failure last time?
• what clippy command should I run?
• what is the project convention for search results?
• what changed in the current product direction?

Run it with:

cargo test --test coding_agent_eval -- --nocapture

The test reports recall@1 and recall@5 and fails if retrieval drops below the stable threshold. This is the product-shaped proof: not broad memory theater, but whether a coding agent can recover the right project context when it matters.

LongMemEval

Retrieval recall on the LongMemEval s_cleaned split — 500 questions over conversational haystacks of ~50 sessions / ~115k tokens each (30 abstention questions are filtered out per the standard convention, leaving 470 evaluated).

The recipe behind the numbers below:

• Granularity: one drawer per session.
• Indexed content: the full session — both user and assistant turns are stored and embedded together. No user-turn filtering, no summarization, no LLM extraction.
• Embedder: all-MiniLM-L6-v2 (384-dim, ONNX), 512-token cap, run locally — no API calls.
• Retrieval: hybrid baseline — BM25 (k1=1.5, b=0.75, weight 0.35) fused with cosine similarity (weight 0.65), top-K = 10. These reported LongMemEval numbers used pure score fusion, before the coding-agent intent boosts used by current project-memory search.
• No LLM at any stage: no extraction, no rerank, no answer generation. The recall numbers measure the retriever in isolation.
• Metric: recall_any@K at session granularity — does any gold session appear in the top-K results?
• Hardware: Apple M1 Pro, 10 cores (8P + 2E), 32 GB RAM.

Split	R@1	R@5	R@10
longmemeval_oracle (sanity check)	1.000	1.000	1.000
longmemeval_s_cleaned	0.889	0.981	0.991

Per-question-type on s_cleaned:

Question type	R@1	R@5	R@10
knowledge-update	0.944	1.000	1.000
multi-session	0.909	0.983	1.000
single-session-assistant	1.000	1.000	1.000
single-session-preference	0.633	0.867	0.933
single-session-user	0.922	1.000	1.000
temporal-reasoning	0.835	0.976	0.984

Reading the numbers

• oracle is a sanity check, not a real result. That split hands the retriever only the sessions known to contain the answer, so perfect recall just confirms the pipeline is wired up correctly.
• s_cleaned is the real test. ~50 sessions / ~115k tokens of conversational haystack per question, no hints. R@5 = 0.981 means that for 461 of 470 evaluated questions, a gold session appears somewhere in the top 5 retrieved.
• R@1 → R@5 → R@10 tells you where the failures cluster. The jump from 0.889 to 0.981 means most "misses" at top-1 are near-misses — the right session is usually rank 2–5, displaced by a lexically similar distractor. The further jump to 0.991 at top-10 means only ~9 questions out of 470 fall outside the top-10 entirely; those are the genuinely hard cases.
• Per-question-type breakdown is where the model's blind spots show.
  • single-session-assistant, single-session-user, knowledge-update: ≥0.94 at R@1, ≈1.0 at R@5. The retriever handles direct questions where the answer is stated verbatim in one session.
  • multi-session and temporal-reasoning: strong at R@5 (~0.98) but lower at R@1 (~0.83–0.91). Multiple sessions are relevant and the "best" one is a judgement call — top-1 ranking among near-equivalents is genuinely ambiguous.
  • single-session-preference: the visible weak spot at 0.633 / 0.867 / 0.933. Preference questions ("what's my favorite X") are answered by sentences like "I like…" / "I prefer…" that don't share keywords with the question. Pure BM25 + frozen MiniLM has no signal for preference-shaped sentences specifically; closing this gap would require either an LLM-extracted preference index or a hand-rolled pattern booster.
• What's deliberately not in these LongMemEval numbers. No LLM at any stage — no extraction during ingest, no query rewriting, no rerank, no answer generation. No per-dataset hyperparameter tuning. No GPU. The result is the baseline retriever in isolation, on a single CPU, with fixed defaults.

---

Installation

macOS / Linux

curl -fsSL https://raw.githubusercontent.com/AncientiCe/mempalace-rs/main/scripts/install.sh | sh

Windows

irm https://raw.githubusercontent.com/AncientiCe/mempalace-rs/main/scripts/install.ps1 | iex

The installer downloads the matching GitHub Release binary, verifies its SHA-256 checksum, installs it locally, and registers the MCP server with Cursor, Codex, and Claude Code.

Development install:

cargo install --path .
mempalace install

The first time you run mine, the embedding model is downloaded automatically from HuggingFace and cached.

---

Quick Start

cargo install --path .       # development install; release installers do this for you
mempalace install            # configures Cursor + Codex + Claude Code
mempalace doctor             # verifies MCP config, rules, binary, and drawer count
mempalace init ~/my-project  # detect rooms and write mempalace.yaml
mempalace mine ~/my-project  # populate the palace

Then restart your agent app so it reloads MCP configuration. Search manually with mempalace search "how did we decide on the database schema" or let your agent call the MCP tools when its installed rule tells it to consult memory.

---

CLI Reference

Command	Description
mempalace init <dir>	Detect rooms from folder structure, write mempalace.yaml
mempalace mine <dir>	Chunk, embed, and store project files
mempalace mine-convos <dir>	Ingest conversation exports
mempalace search <query>	Semantic search with similarity scores
mempalace wake-up	Print L0 (identity) + L1 (essential story) context
mempalace status	Palace overview: drawer counts by wing/room
mempalace gain	Show MCP usage gains, estimated savings, and per-project value
mempalace split	Split Claude Code mega-transcripts by session
mempalace repair	Re-embed any drawers missing vectors
mempalace install	Register the MCP server with Cursor, Codex, and Claude Code
mempalace uninstall	Remove MemPalace from MCP client configs
mempalace doctor	Inspect binary path, palace DB, and MCP config status
mempalace mcp	Start the MCP stdio server

mine flags

mempalace mine ~/my-project \
  --wing my_project          # Override wing name
  --limit 100                # Cap at 100 files
  --dry-run                  # Preview without storing
  --no-gitignore             # Ignore .gitignore rules
  --include vendor,third_party  # Force-include these paths

mine-convos flags

mempalace mine-convos ~/Desktop/transcripts \
  --wing claude_sessions \
  --mode exchange   # or: general (decisions/milestones/emotions)
  --limit 50
  --dry-run

split flags

mempalace split \
  --source ~/Desktop/transcripts \
  --min-sessions 2 \
  --dry-run

gain

mempalace gain summarizes automatic MCP usage by Cursor, Codex, Claude Code, or any other MCP client. It records local tool-call metadata in palace.db and estimates value from retrieval hits, duplicate skips, KG facts, diary recalls, repeat questions, and latency.

mempalace gain
mempalace gain --project my_project --since 7d
mempalace gain --history
mempalace gain --json

Example output:

MemPalace gain - last 30d (mempalace_rs)
  Tool calls         : 412   (sessions: 27)
  Hit rate           : 88%   (search hits 142/162)
  Tokens saved (est) : ~78,400
  Re-index skipped   : 31    (duplicate drawers avoided)
  KG facts recalled  : 56
  Diary recalls      : 8
  Repeat Qs avoided  : 19
  p95 latency        : 41 ms
  Top wings          : mempalace_rs(120), checkout(40)

Set MEMPALACE_GAIN_DISABLED=1 to disable usage recording.

---

MCP Setup

mempalace install is the normal setup command for all three major agentic coding clients. It writes both:

• an MCP server entry that starts mempalace mcp
• a small rule that tells the agent when to call mempalace_status, mempalace_search, mempalace_kg_query, and mempalace_diary_write

mempalace install

What gets written by default:

Client	MCP config	Rule file
Cursor	~/.cursor/mcp.json	~/.cursor/rules/mempalace.mdc
Codex	~/.codex/config.toml	~/.codex/AGENTS.md
Claude Code	~/.claude/mcp_servers.json	~/.claude/CLAUDE.md

Install for one client:

mempalace install --client cursor
mempalace install --client codex
mempalace install --client claude

Install project-scoped rules instead of global rules:

mempalace install --scope project --path /path/to/project

For project scope, Cursor also gets a project-local MCP config at <project>/.cursor/mcp.json. Codex and Claude Code keep MCP config in their user-level config files, while their rules go into <project>/AGENTS.md and <project>/CLAUDE.md.

Skip rule files if you only want MCP wiring:

mempalace install --no-rule

Inspect the current setup:

mempalace doctor

Remove MemPalace config:

mempalace uninstall
mempalace uninstall --client cursor

Cursor

After mempalace install --client cursor, restart Cursor or reload the window. Settings -> MCP should show mempalace as an enabled stdio server.

Manual Cursor config shape:

{
  "mcpServers": {
    "mempalace": {
      "command": "mempalace",
      "args": ["mcp"]
    }
  }
}

The rule is installed as .cursor/rules/mempalace.mdc with alwaysApply: true.

Codex

After mempalace install --client codex, restart Codex so it reloads ~/.codex/config.toml.

Manual Codex config shape:

[mcp_servers.mempalace]
command = "mempalace"
args = ["mcp"]

The rule is installed as a managed MemPalace block in ~/.codex/AGENTS.md (or <project>/AGENTS.md with --scope project). Existing content is preserved.

Claude Code

After mempalace install --client claude, restart Claude Code so it reloads ~/.claude/mcp_servers.json.

Manual Claude JSON shape is the same as Cursor's mcpServers object above. You can also use Claude Code's own MCP command:

claude mcp remove mempalace
claude mcp add mempalace -- mempalace mcp

The rule is installed as a managed MemPalace block in ~/.claude/CLAUDE.md (or <project>/CLAUDE.md with --scope project). Existing content is preserved.

MCP Tools

The server exposes tools for status, taxonomy, search, drawer CRUD, knowledge graph operations, graph tunnels, hook acknowledgements, and agent diaries:

Tool	Description
mempalace_status	Palace overview + protocol
mempalace_gain	MCP usage gains, estimated savings, and per-project value
mempalace_list_wings	List wings with drawer counts
mempalace_list_rooms	List rooms within a wing
mempalace_get_taxonomy	Full wing → room → count tree
mempalace_get_aaak_spec	AAAK compressed memory dialect spec
mempalace_search	Semantic search over drawers
mempalace_check_duplicate	Check if content already exists
mempalace_add_drawer	File content into the palace
mempalace_delete_drawer	Remove a drawer by ID
mempalace_kg_query	Query entity relationships
mempalace_kg_add	Add a fact (subject → predicate → object)
mempalace_kg_invalidate	Mark a fact as no longer true
mempalace_kg_timeline	Chronological fact history
mempalace_kg_stats	Knowledge graph overview
mempalace_traverse	BFS graph walk from a room
mempalace_find_tunnels	Rooms bridging two wings
mempalace_graph_stats	Palace graph summary
mempalace_diary_write	Write a diary entry in AAAK format
mempalace_diary_read	Read recent diary entries

---

Migration from Python

The Rust version uses a new single-file database (palace.db). Your existing ChromaDB data cannot be migrated automatically.

Steps:

# 1. Re-mine your projects
mempalace init ~/my-project && mempalace mine ~/my-project

# 2. Re-index conversations
mempalace mine-convos ~/Desktop/transcripts

# 3. Verify
mempalace status

Your identity.txt, people_map.json, and known_names.json in ~/.mempalace/ are compatible and will be read automatically.

---

Test on a Project

mempalace init /path/to/project
mempalace mine /path/to/project
mempalace status

Restart Cursor, Codex, or Claude Code, then ask the agent a project question that should use memory, for example: "Search MemPalace for how this project handles database migrations." The agent should call mempalace_search through MCP instead of re-indexing the repository from scratch.

---

Configuration

~/.mempalace/config.json is read on startup. Environment variables take highest priority:

Env Var	Default	Description
MEMPALACE_PALACE_PATH	~/.mempalace/palace	Palace data directory

mempalace.yaml (per-project)

Created by mempalace init. Example:

wing: my_project
rooms:
  - name: backend
    description: Server and API code
    keywords: [api, server, routes, models]
  - name: frontend
    description: UI components
    keywords: [ui, components, pages, views]
  - name: general
    description: Everything else
    keywords: []

---

Memory Stack

Layer	Name	Description
L0	Identity	~/.mempalace/identity.txt — always loaded (~100 tokens)
L1	Essential Story	Top drawers by importance, grouped by room (~600–900 tokens)
L2	On-Demand	Wing/room filtered retrieval
L3	Deep Search	Full semantic search

mempalace wake-up prints L0 + L1. The AI uses MCP tools for L2/L3.

---

Development

cargo build
cargo test
cargo clippy

Tests use in-memory SQLite — no palace.db needed. The embedding model is not loaded in tests that don't require it.

---

Hooks Compatibility

Shell hooks that previously called python -m mempalace.mcp_server can now call mempalace mcp. Update the binary path in your hooks:

# Before (Python)
exec python -m mempalace.mcp_server

# After (Rust)
exec mempalace mcp