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/rtk-ai/icm/agents-mdgit clone --depth 1 https://github.com/rtk-ai/icmWhat 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 | $0.00357 | $0.00357 |
| Opus 5 | $0.00179 | $0.00179 |
| Sonnet 5 | $0.00071 | $0.00071 |
| Haiku 4.5 | $0.00036 | $0.00036 |
Grade A, and why
icm 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 2d 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.
What it actually says
Persistent memory (ICM) — MANDATORY
This project uses ICM for persistent memory across sessions. You MUST use it actively. Not optional.
Recall (before starting work)
icm recall "query" # search memories
icm recall "query" -t "topic-name" # filter by topic
icm recall-context "query" --limit 5 # formatted for prompt injection
Store — MANDATORY triggers
You MUST call icm store when ANY of the following happens:
- Error resolved →
icm store -t errors-resolved -c "description" -i high -k "keyword1,keyword2" - Architecture/design decision →
icm store -t decisions-{project} -c "description" -i high - User preference discovered →
icm store -t preferences -c "description" -i critical - Significant task completed →
icm store -t context-{project} -c "summary of work done" -i high - Conversation exceeds ~20 tool calls without a store → store a progress summary
Do this BEFORE responding to the user. Not after. Not later. Immediately.
Do NOT store: trivial details, info already in CLAUDE.md, ephemeral state (build logs, git status).
Other commands
icm update <id> -c "updated content" # edit memory in-place
icm health # topic hygiene audit
icm topics # list all topics
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.
- 2d ago First seen · 33 lines · 357 tokens per session scan A 73e0a225e0fd
icm AGENTS.md is an instructions file published in the GitHub repository rtk-ai/icm (549 stars, last pushed 3d ago), licensed Apache-2.0. It adds 357 tokens to every session, about $0.0018 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
grit AGENTS.md
Instructions for rtk-ai/grit, covering persistent memory (icm), recall (before starting work), store (after completing significant work) and other commands.
grit CLAUDE.md
Instructions for rtk-ai/grit, covering persistent memory (icm), recall (before starting work), store (after completing significant work), other commands and rtk (rust token killer) - token-optimized commands.
agentgateway copilot-instructions.md
Copilot instructions for agentgateway/agentgateway: Do not check for, speculate about, or report compilation errors during code review. Compilation diagnostics from review are frequently incorrect; rely on CI to detect and report compilation failures.
mq commit.instructions.md
Instructions for harehare/mq: Use the following format for commit messages.
playwright-rust CLAUDE.md
Instructions for padamson/playwright-rust, covering claude.md, project, repository layout, skills (procedural reference) and documentation hierarchy.
codemark GEMINI.md
Instructions for DanielCardonaRojas/codemark, covering development workflow and debugging with the tui logging system.