What is lethe AGENTS.md?

Project instructions for lethe, a self-improving memory store for AI agents. Lethe indexes agent transcripts and searches them using both keyword and meaning-based matching, with a Rust command-line program and language bindings.

Use it when working on lethe's Rust code, command-line interface, Python or Node.js bindings, memory storage, tests, or repository hooks.

It explains the project's structure, tools, storage, and checks so an agent can make changes consistently in this Rust workspace.

lethe AGENTS.md — Instructions file for Codex and OpenCode by teimurjan: Project instructions for lethe, a self-improving memory store for…

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods

npx agentmods add instructions/teimurjan/lethe/agents-md

Clone the repo

git clone --depth 1 https://github.com/teimurjan/lethe

Made for: Codex, OpenCode.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

README.md

[![agentmods](https://agentmods.dev/badge/instructions/teimurjan/lethe/agents-md.svg)](https://agentmods.dev/instructions/teimurjan/lethe/agents-md)

Your own site

<a href="https://agentmods.dev/instructions/teimurjan/lethe/agents-md"><img src="https://agentmods.dev/badge/instructions/teimurjan/lethe/agents-md.svg" alt="Measured on agentmods" height="20"></a>

Per session 2,330 This file is loaded in full into every session.

When invoked 2,330 The same file — it is already loaded in full.

Security scan A 0 findings. Scan, not verified.

Origin original No closer match found in the catalogue.

Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

Model	Per session	Once invoked
Fable 5	$0.02330	$0.02330
Opus 5	$0.01165	$0.01165
Sonnet 5	$0.00466	$0.00466
Haiku 4.5	$0.00233	$0.00233

Measured 3d ago against content hash b1797bb65702, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

lethe AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 176 lines

How it starts

The opening of the file, as written. The whole thing — 176 lines — stays where its author put it; the contents beside it link to each section on GitHub.

lethe — agent guide

Self-improving memory store for LLM agents. BM25 + dense vector hybrid retrieval with cross-encoder reranking and clustered retrieval-induced forgetting (RIF). Indexes agent transcripts (Claude Code / Codex) directly — no capture hooks, no summarization. Ships as a single Rust binary (lethe) plus PyPI / npm bindings.

stack

Rust workspace (crates/) — production. Toolchain pinned via rust-toolchain.toml.
- lethe-core — library: tokenize, bm25, faiss_flat, rrf, dedup, rif, kmeans, encoders (ONNX via ort), DuckDB persistence, npz reader, memory_store, union_store, markdown_store, transcript_store.
- lethe-cli — lethe binary (clap; embeds the TUI).
- lethe-tui — ratatui library called from the CLI on no-arg invocation.
- lethe-py — PyO3 binding → PyPI lethe-memory.
- lethe-node — napi-rs binding → npm lethe.
- lethe-benchmark — internal parity bench helper (publish = false).
DuckDB for entry metadata + embedding BLOBs (single source of truth).
ONNX Runtime (via ort) for the bi-encoder + cross-encoder. Default models: Xenova/all-MiniLM-L6-v2, Xenova/ms-marco-MiniLM-L-6-v2.
Python research_playground/lethe_reference/ kept for the research-journey trail: produces the benchmark numbers cited below. Installed as lethe-memory-legacy in dev venvs; not published.
Datasets (tmp_data/, gitignored): LongMemEval (S), NFCorpus.

commands

# Rust workspace
cargo build --workspace --release        # → target/release/lethe
cargo test --workspace
cargo clippy --workspace --all-targets -- -D warnings
cargo fmt --all -- --check

# Pre-commit / pre-push hooks (replaces the deleted ci.yml)
cargo install --git https://github.com/j178/prek
prek install                             # writes .git/hooks/pre-commit + pre-push
prek run --all-files                     # one-off run

# Python reference impl (research trail; library only)
uv pip install -e research_playground/lethe_reference/
cd research_playground/lethe_reference && uv run pytest tests/ -v     # 148 + 8 PyO3 parity = 156
uv run python research_playground/baseline/run.py

# Python ↔ Rust parity bench
uv run python research_playground/rust_migration/prepare.py
uv run python research_playground/rust_migration/longmemeval.py --compare
uv run python research_playground/rust_migration/components.py --compare
uv run python research_playground/rust_migration/latency.py --compare

# CLI surface (the `lethe` binary)
lethe                                    # no args → TUI (in a terminal)
lethe index                              # index this project's transcripts
lethe index --all                        # reindex every registered project
lethe search "query" --top-k 5           # reindexes changed transcripts first
lethe search "query" --all --top-k 5     # all registered projects (per-project read-only)
lethe projects list|add|remove|prune
lethe expand <chunk-id> [<chunk-id> ...]
lethe status

Read the full file on GitHub · 176 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

3d ago First seen · 176 lines · 2,330 tokens per session scan A b1797bb65702

Subscribe to this mod's changes

lethe AGENTS.md is an instructions file published in the GitHub repository teimurjan/lethe (11 stars, last pushed 1mo ago), licensed MIT. It adds 2,330 tokens to every session, about $0.0117 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.

LLM agents bm25 duckdb

lethe AGENTS.md

Getting it into your agent

Wrote this? Show the measurements

What it costs to keep this loaded

Grade A, and why

Nothing flagged

How it starts

lethe — agent guide

stack

commands

What this file has done since we first saw it

Other instructions, from other repositories

WrenAI AGENTS.md

tokenjam CLAUDE.md

tokenjam AGENTS.md

OpenNotebookLM AGENTS.md

Engram AGENTS.md

youtube-advisor AGENTS.md