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/tt-wang/memem/claude-mdgit clone --depth 1 https://github.com/TT-Wang/mememWrote 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/tt-wang/memem/claude-md)<a href="https://agentmods.dev/instructions/tt-wang/memem/claude-md"><img src="https://agentmods.dev/badge/instructions/tt-wang/memem/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 | $0.03907 | $0.03907 |
| Opus 5 | $0.01954 | $0.01954 |
| Sonnet 5 | $0.00781 | $0.00781 |
| Haiku 4.5 | $0.00391 | $0.00391 |
Grade A, and why
memem 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 4d 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.
How it starts
The opening of the file, as written. The whole thing — 198 lines — stays where its author put it; the contents beside it link to each section on GitHub.
memem — Persistent Memory & Context Assembly
You have access to a persistent memory system via memem. This memory persists across sessions, self-evolves, and gets smarter over time.
Storage: Obsidian vault at ~/obsidian-brain/memem/memories/ (markdown files, single source of truth).
Machine index: SQLite FTS5 at ~/.memem/search.db for fast retrieval.
Auto-recall
In auto mode, the UserPromptSubmit hook fires on every message and builds an active_memory_slice from the current query. As of v2.4.0, tool is the default mode — auto-injection is off; Claude pulls memory on demand via the MCP tools below.
For on-demand recall, use the MCP tools below.
Context model (v2.8+)
v2.8.0 replaces the L0–L3 layer system with a three-tier context model. Layers are retired — the data falsified them.
Tier 1 — Profiles (always-injected)
Profile documents at <vault>/memem/profiles/ are injected at every session start before any memory recall occurs. Two document types:
profile_user.md— sections: Preferences / Conventions / Environmentprofile_<project>.md— sections: Identity / Stack & Structure / Conventions
Profiles are NOT in the retrieval corpus (you will never see them in memory_search results). Their consumer is the SessionStart hook. The miner writes to profiles via the PROFILE reconcile op; you can also populate them via --migrate-layers --apply.
Tier 2 — Working rules (procedural, citation-ranked)
At session start you receive a ## Working rules block containing type:procedural memories — failure→fix patterns and correction knowledge — ranked by citation count (last 30 days) then by recency, capped to ≤1200 chars. These are the rules you have cited most recently in your work.
Tier 3 — Episode index + on-demand retrieval
At session start you also receive a ## Episode index of up to 25 recent type:episodic memory titles. Everything else is available on demand via the recall tools below.
During the session, use the 3-tier recall workflow:
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.
- 4d ago First seen · 198 lines · 3,907 tokens per session scan A 8706cf63d492
memem CLAUDE.md is an instructions file published in the GitHub repository TT-Wang/memem (32 stars, last pushed 2d ago), licensed MIT. It adds 3,907 tokens to every session, about $0.0195 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.
Other instructions, from other repositories
captain-memo AGENTS.md
Instructions for kalinbogatzevski/captain-memo, covering agents.md, what this project is, runtime, commands and layout.
neurostack CLAUDE.md
Instructions for raphasouthall/neurostack, covering neurostack - claude code guide, quick reference, installation, mcp server (recommended for claude code) and openai-compatible api.
rag-rat AGENTS.md
Instructions for cq27-dev/rag-rat, covering agents.md, prefer the rag-rat mcp for code browsing, record durable learnings as rag-rat memories, repo orientation and worktree correctness.
wenlan CLAUDE.md
Claude Code instructions for 7xuanlu/wenlan, a project described as: Wenlan is a knowledge base for the AI-native age. Your AI agents capture what they learn, Wenlan keeps it current and distills it into source-cited wiki pages you can trust.
MCP-memento copilot-instructions.md
Instructions for caiowilson/MCP-memento, covering mcp tooling policy (vs code), project architecture (big picture), critical workflows, project-specific conventions and tool notes.
delx-memory AGENTS.md
AGENTS.md instructions for davidmosiah/delx-memory, covering agent development notes, scope, commands, rules and agent-readiness checklist.