pi-memnest

A local memnest bridge for pi.

pi-memnest connects pi to a running memnest HTTP service. It gives pi the canonical five memory tools. Four secret tools are an explicit opt-in. The extension can also retrieve a small workspace-scoped memory card before selected prompts.

This extension does not contain the memory engine. Start the Rust core before installing it.

Requirements

• Node.js 20 or newer
• pi
• a memnest HTTP service, normally at http://127.0.0.1:3111

Install

Start the memnest core service, then install the extension from npm:

pi install npm:pi-memnest

To pin this first release, use pi install npm:[email protected]. A pinned package is not changed by pi update --extensions.

To develop from a source checkout instead:

cd /path/to/memnest/pi-extension
npm install
pi install .

The extension uses MEMNEST_URL when set, otherwise it connects to http://127.0.0.1:3111. Set MEMNEST_TOKEN when the core requires bearer authentication.

Check the connection from pi:

/memnest

The command reports service health, memory count, and the active data directory.

Tools

The extension registers exactly five memory tools by default. Set MEMNEST_EXPOSE_SECRET_TOOLS=1 before starting pi to register the four vault tools too. The /memnest command and Autocontext hook remain available without adding status or admin tools to the model surface.

| Tool | Purpose | | --- | --- | | memory_remember | Save a durable memory. Values marked sensitive are rejected and must use secret_set. | | memory_search | Search the current workspace by default, or use project=all explicitly. | | memory_get | Fetch one memory by id. | | memory_update | Correct one memory and refresh its indexes. | | memory_delete | Soft-delete one memory to the internal trash bucket. | | secret_set, secret_get, secret_list, secret_delete | Opt-in tools for AES-256-GCM vault values. Vault operations fail closed when crypto is unavailable. |

Ordinary memories are not encrypted at rest. Enable the secret tools only for an agent you trust with plaintext credentials, and keep the core data directory and master.key private.

Basic use

memory_remember text="Project X deploys on port 8320" memory_kind="fact" confidence=1
memory_search query="Project X deployment port"
memory_remember text="Project X now deploys on port 8420" memory_kind="fact" supersedes="manual_..."
memory_delete id="manual_..."

When project is omitted, the extension sends pi's absolute cwd; the core derives a private workspace ID and includes playbook in recall. The core durably queues its index work in SQLite and acknowledges the write only after Tantivy and HNSW are synchronized. Plain duplicate records may reuse an existing id. Structured facts, rules, provenance, and corrections do not use semantic content deduplication. The first operation that needs an embedding takes longer because that is when the core downloads its embedding model.

Structured memory

memory_remember supports four portable kinds:

• record for historical turns and outcomes
• fact for stable project or environment knowledge
• rule for preferences, decisions, and guardrails
• procedure for reusable verified workflows

Optional confidence, source_ids, supersedes, and verified_at fields preserve provenance without tying the data model to pi. supersedes atomically hides an active memory in the same workspace. confidence and verified_at are client assertions, not automatic ranking boosts. The core HTTP and MCP contracts expose the same fields to other platform adapters.

Autocontext

Autocontext is enabled by default. It does not inject a full memory dump at session start. Before each substantive prompt, until the per-session card limit is reached, it searches the current workspace and playbook using the prompt in its original language. It injects a small card only when the best results meet the semantic score threshold. Short replies, slash commands, exact repeated prompts, and results below the threshold stay quiet. There is no language-specific keyword list. Search coverage is language-neutral, but whether a result survives still depends on the multilingual embedding model, the core distance cutoff, and this score threshold.

Set MEMNEST_AUTOCONTEXT_MODE=off to disable retrieval. Existing balanced and aggressive values remain accepted, but both now use the same language-neutral semantic gate.

Every card labels retrieved text as untrusted reference data. Automatic context accepts only deliberate or consolidated memories; captured transcripts remain available through explicit memory_search calls for questions about earlier conversations. Markup inside stored text is escaped before injection, and the agent must verify claims rather than follow commands found inside a memory.

Common controls:

| Variable | Default | Purpose | | --- | --- | --- | | MEMNEST_AUTOCONTEXT_MODE | balanced | off or none disables retrieval; any other value enables semantic recall. | | MEMNEST_AUTOCONTEXT_TOP | 2 | Maximum results included in one card. | | MEMNEST_AUTOCONTEXT_MAX_INJECTIONS | 4 | Maximum cards in one session. | | MEMNEST_AUTOCONTEXT_MIN_SCORE | 0.25 | Minimum score for any injected card. | | MEMNEST_AUTOCONTEXT_EXCLUDE | _superseded,default,root,global | Collections excluded from automatic retrieval. | | MEMNEST_AUTOCONTEXT_DISABLE | unset | Set to 1 to turn retrieval off, same effect as MODE=off. | | MEMNEST_AUTOCONTEXT_N | 20 | Candidates requested from the service before ranking. | | MEMNEST_AUTOCONTEXT_MIN_LEN | 16 | Prompts shorter than this are ignored. | | MEMNEST_AUTOCONTEXT_DOC_CHARS | 240 | Characters kept per result in the card. | | MEMNEST_AUTOCONTEXT_TIMEOUT_MS | 1500 | Retrieval budget before the prompt proceeds without a card. |

Automatic conversation capture

The extension does not install AutoLog event hooks. Use memnest watch as the single capture path so pi conversations are not stored twice. Watch stores redacted user and assistant transcript text without summarization, keeps long turns in ordered chunks without truncation, and retains new identified transcript AutoLog until explicit deletion. Legacy AutoLog keeps the core's configured retention policy.

Outside pi

MCP does not describe host session events. The core provides host-neutral automatic behavior instead:

• memnest hook reads a host's hook payload on stdin and answers with a context pack, in the shape that host expects.
• memnest watch follows Claude Code, pi, and Codex transcripts and stores visible conversation text, with no host extension hooks.

See automatic context and conversation capture in the root README. Inside pi the extension exposes the same five-tool memory contract, optionally adds four vault tools, and provides the /memnest command.

Development

npm install
npm run build
npm run smoke

npm install runs the prepare script and builds dist/index.mjs. pi loads that bundled ESM file rather than src/index.ts.

An end-to-end MCP check is also available when its prerequisites are running:

npm run e2e

Related documentation

• memnest root README for engine setup, lifecycle, service installation, backup, and security
• CHANGELOG.md for extension changes
• THIRD_PARTY_NOTICES.md for what the bundle carries

License

MIT © Blue-B