The attack is a sleeping sentence your logs cannot see

Local-first memory for AI agents that finds the cause, not just the match.

Vestige remembers your decisions, catches contradictions before they cost you, and traces a failure back to the older memory that actually caused it. One 25MB Rust binary over MCP. No cloud, no API keys, no telemetry. Your data never leaves your machine.

Install · Why not RAG · Benchmark · Science · Tools · Dashboard · Pro · Docs

Agents re-learn the same lessons: they recommend a change you already tested and rejected, re-derive a fix that was already written down, and treat every session as if the last one never happened. Vestige is the memory layer that ends that. Any MCP-capable agent (Claude Code, Claude Desktop, Codex, Cursor, and others) writes memories as you work and retrieves them later, modeled on real cognitive science: redundant memories merge, contradicted ones are flagged, unused ones fade, and when a failure hits, Vestige reaches backward to the decision that set it up.

The cause never looks like the bug. That is the whole product.

You need Node.js. No Docker, no signup, no compile step (prebuilt for macOS ARM + Intel, Linux x86_64, Windows x86_64).

npm install -g vestige-mcp-server@latest

Connect it to your agent. Every MCP client understands this config:

{
  "mcpServers": {
    "vestige": { "command": "vestige-mcp" }
  }
}

docs/CONFIGURATION.mdVerify: vestige dashboard

, then open ** http://localhost:3927/dashboard**. First run downloads a 130MB embedding model once; after that Vestige is fully offline, forever. Full walkthrough:

docs/GETTING-STARTED.md.

RAG retrieves text that resembles the query. That is the right tool when the answer looks like the question, and the wrong tool when the cause of a problem looks nothing like the symptom: a config choice from three weeks ago, a library pin, an assumption nobody flagged as risky.

The backward reach implements Retroactive Salience Backfill (Zaki, Cai et al., Nature 2024, 637:145-155, DOI 10.1038/s41586-024-08168-4): when a memory turns out to matter, the salience of the earlier memories that led to it is raised, so the causal chain becomes retrievable even though the surface text never matched. Every backfill result ships with a receipt naming the exact evidence path; Vestige reports receipt-backed candidate causes, never an unverifiable verdict.

And the limitation on the left column is not marketing: DeepMind proved single-vector retrieval mathematically incapable of certain relevance patterns (arXiv:2508.21038, ICLR 2026).

The claim is testable, and the test ships with all 246 agent transcripts it produced. Three coding agents fix one failing e2e test; the fix needs the currently live signing key id, randomized per trial from a 50-key keyring, present in no file the agents can read. It exists only in the memory layer. The dangerous outcome is converging on a planted decoy: tests pass, the merge is clean, production breaks.

| Arm (6 models, 25 trials) | Converged correct | Converged wrong | Split | |---|---|---|---| | No memory | 0/25 | 21/25 | 4/25 | | Dense cosine RAG | 4/23 | 12/23 | 7/23 | | Vestige | 20/23 | 0/23 | 3/23 |

On the verbatim queries the agents typed, the causal memory ranks 7th of 8 under both dense cosine and BM25 while the decoy ranks 1st. Reproduce the central measurement in two seconds, stdlib only:

git clone -b benchmark/silent-rotation --depth 1 https://github.com/samvallad33/vestige.git
cd vestige/benchmarks/silent-rotation
python3 tests/bm25_baseline.py results/runA-trial-1/corpus-export.json --no-dense

The caveats are published alongside the results, including the trials a plain cosine baseline ties and the trial Vestige loses.

Every mechanism is a cited result, implemented in Rust, running locally. Full write-up: docs/SCIENCE.md.

Your agent calls these; you rarely do.

memory

· graph

· intention

maintain

· dedup

· suppress

memory_status

· codebase

· source_sync

· session_start

Project scoping, hygiene workflows, and making memory a standing habit for your agent: docs/MEMORY_HYGIENE.md · docs/AGENT-MEMORY-PROTOCOL.md · docs/CLAUDE-SETUP.md.

vestige dashboard

A living WebGPU observatory of your memory at ** http://localhost:3927/dashboard**: memories appear, link, strengthen, and fade in real time, 1000+ nodes at 60fps. It renders a deterministic 12-second loop of your store's life that you can export as an mp4 with one click, and mints a

brain print, a signature seeded from your store's shape. Share artifacts are structure-only by design: your brain, never your memories.

Everything above is free forever and never metered. Pro ($19/month) is managed, end-to-end encrypted continuity: your memory graph and accountability history (receipts, traces, memory PRs) following you across machines. XChaCha20-Poly1305 applied on your device, Argon2id over a passphrase only you know, ciphertext-only server. Zero-knowledge is the design: lose the passphrase and the data is unrecoverable, by anyone. Checkout opens shortly; watch Releases for the announcement.

| Engine | Rust 2024, ~96k lines, single 25MB binary, 1,961 tests, clippy clean at -D warnings | | Retrieval | Nomic Embed v1.5 (Matryoshka 768d→256d) + USearch HNSW + SQLite FTS5, optional Qwen3 reranker | | Storage | SQLite, optional SQLCipher encryption ( |

Getting Started · FAQ · The Science · Configuration · Storage · Silent Rotation · Changelog

If Vestige saves you from one repeated mistake, that is the whole point: never solve the same problem twice. If it earns a place in your setup, a star genuinely helps.