memgit CLAUDE.md

A set of working instructions for memgit, a tool that gives AI agents a durable memory store managed with Git. It covers releases, backups, search naming, and how to record testing findings.

In plain words
What is it for?
Use it when maintaining memgit, publishing its updates, backing up its memory store, or recording problems found while using it.
Why use it?
It helps keep the tool's stored memories, releases, and maintenance rules consistent across sessions.

Instructions file

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/code4161/memgit/claude-md
Clone the repo
git clone --depth 1 https://github.com/code4161/memgit
Per session 575 This file is loaded in full into every session.
When invoked 575 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.

ModelPer sessionOnce invoked
Fable 5 $0.00575 $0.00575
Opus 5 $0.00287 $0.00287
Sonnet 5 $0.00115 $0.00115
Haiku 4.5 $0.00057 $0.00057

Measured today against content hash e7b95217fb0d, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

memgit CLAUDE.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 today.

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.

CLAUDE.md · 38 lines

How it starts

The opening of the file, as written. The whole thing — 38 lines — stays where its author put it; the contents beside it link to each section on GitHub.

memgit core — operating guide

Git for AI memory: Python CLI + MCP server + hooks + VS Code extension (repo code4161/memgit, public). The live store on this machine is ~/.claude/memgit-store/; the installed tool runs from the system Python 3.13 site-packages — this checkout is not the live tool.

Rules

  • Release via the memgit-maintenance skill — tag push auto-publishes PyPI/npm/choco; brew tap, vsce publish and the Claude plugin are manual, and README/docs must move in the same release. The 2026-08-05 audit traced nearly every live defect to ONE unreleased commit: a merged feature is dead until released and installed.
  • Autonomy is a design constraint (memgit-autonomous-durability): a maintenance task requiring a human command will never happen. Backup is automatic end-of-session; memgit never invents a git remote.
  • MCP registry search matches NAME ONLY (mcp-registry-search-is-name-only…) — naming decisions are distribution decisions and effectively permanent.
  • Dogfooding findings become memories (project Personal-business) AND GitHub issues — the store this tool manages is also its best test fixture.
  • Pricing/billing decisions: memgit-cloud-billing-august-cash + MEMGIT_NEXT_STEPS gates. No money on memgit-cloud until the free test converts.

Doc: MEMGIT_NEXT_STEPS.md. Group map: ../CLAUDE.md.

🔴 CORE RULE — no conflicting or stale memory (workspace-wide, 2026-08-31)

save_memory is not append-only in effect: a stale memory competes with its own correction and can OUTRANK it in search. Before every save — search_memories on the QUANTITY, and if an existing memory states a different value, re-save that one or pass supersedes. Correct the rule line, not just the body (search and digests render the rule). If a correction inverts a finding, the slug changes too. lx lessons carry no operational directives ("do not arm") — those expire with state, and state lives in the tr tracker. save_memory on an existing slug REPLACES: always resend the full body. Close the loop — re-search what you wrote and read the top hit.

Read the full file on GitHub · 38 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.

  1. today First seen · 38 lines · 575 tokens per session scan A e7b95217fb0d

Subscribe to this mod's changes

memgit CLAUDE.md is an instructions file published in the GitHub repository code4161/memgit (0 stars, last pushed yesterday), licensed MIT. It adds 575 tokens to every session, about $0.0029 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 2026-08-31.