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.
npx agentmods add instructions/jagoff/memo/claude-mdgit clone --depth 1 https://github.com/jagoff/memoWrote 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.
[](https://agentmods.dev/instructions/jagoff/memo/claude-md)<a href="https://agentmods.dev/instructions/jagoff/memo/claude-md"><img src="https://agentmods.dev/badge/instructions/jagoff/memo/claude-md.svg" alt="Measured on agentmods" height="20"></a>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 | $0.09483 | $0.09483 |
| Opus 5 | $0.04741 | $0.04741 |
| Sonnet 5 | $0.01897 | $0.01897 |
| Haiku 4.5 | $0.00948 | $0.00948 |
Grade B, and why
memo CLAUDE.md scanned grade B with 1 finding 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 5d 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.
Reads agent configuration directoriesmediumAgent snooping
.claude/, .codex/, .gemini/ hold keys, settings and other credentials a mod has no legitimate need for.
entry into `~/.claude/settings.json` — NOT `hooks/hooks.json`, which owns the How it starts
The opening of the file, as written. The whole thing — 583 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md
This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.
memo is a local-first semantic memory: MLX embeddings + sqlite-vec hybrid
(vec + BM25) search, reranking, a knowledge graph, temporal reasoning, an MCP
server, and a CLI. Single-user, runs offline. PyPI distribution name is mlx-memo; CLI binary is memo; MCP server binary is memo-mcp.
Dev commands
# One-time setup (fresh clone/worktree — pytest is an optional-dependency
# extra, not a base dep, so plain `uv sync` alone leaves pytest missing)
uv sync --extra dev # pytest, ruff, mypy, hypothesis, etc.
uv sync --all-extras # + http/tantivy/multimodal/ocr, needed for those test modules
# Tests (no MLX needed — MLX tests skip automatically on non-Apple Silicon)
uv run --no-sync pytest tests/ # full suite
uv run --no-sync pytest tests/test_foo.py::test_bar -v # single test
uv run --no-sync pytest tests/ -m "not slow" # skip real-MLX-load tests (>1s each)
# Type checking
uv run --no-sync mypy src/memo/
# Lint + format
uv run --no-sync ruff check src/ && uv run --no-sync ruff format src/
# Run the CLI
uv run --no-sync memo <cmd>
# Validate MEMO_* flags (typos, unknown vars)
uv run --no-sync memo config validate
If memo doctor warns memo Python package loaded from .../src/memo/..., outside the isolated runtime, a stray PYTHONPATH=src (relative to a ~/repos/memo
cwd) is shadowing the isolated uv tool install with this working tree's
source — memo --version looks right while the code that actually runs is
stale. unset PYTHONPATH before verifying a fix against the real installed
binary.
Working tree is shared (read before any git op)
This repo is frequently worked on by concurrent agent sessions sharing one
working tree and HEAD. Their commits land on whatever branch is checked out,
master advances underneath you, and a git checkout in another session moves
your HEAD too. So:
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.
- 5d ago First seen · 583 lines · 9,483 tokens per session scan B 179a53dfa43c
memo CLAUDE.md is an instructions file published in the GitHub repository jagoff/memo (17 stars, last pushed today), licensed MIT. It adds 9,483 tokens to every session, about $0.0474 per session on Opus 5. A static security scan graded it B with 1 finding (reads agent configuration directories). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
speech-swift AGENTS.md
AGENTS.md instructions for soniqo/speech-swift, covering agent instructions, communication style, workflow, running tests and benchmarks — sequential and memory-aware and git conventions.
flyto-core AGENTS.md
AGENTS.md instructions for flytohub/flyto-core, covering flyto-core agent rules, flyto2 project memory contract, flyto2 frontend quality gate and repo notes.
speech-studio AGENTS.md
Instructions for soniqo/speech-studio, covering agent instructions, project, stack, sibling repos under /repos/ and build.
OpenSuperMLX AGENTS.md
AGENTS.md instructions for axot/OpenSuperMLX, covering agents.md — opensupermlx, agent behavioral guidelines, 1. think before coding, 2. simplicity first and 3. surgical changes.
alcove AGENTS.md
AGENTS.md instructions for epicsagas/alcove, covering claude.md, repository purpose, actual location, structure and karpathy 3-layer architecture.
swag-mcp CLAUDE.md
Claude Code instructions for jmagar/swag-mcp, covering swag mcp - claude memory reference, project overview, core architecture, key components and common development commands.