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/coldlogicai/3notch/agents-mdgit clone --depth 1 https://github.com/coldlogicAI/3notchWhat 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 | $0.01527 | $0.01527 |
| Opus 5 | $0.00763 | $0.00763 |
| Sonnet 5 | $0.00305 | $0.00305 |
| Haiku 4.5 | $0.00153 | $0.00153 |
Grade A, and why
3notch 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.
How it starts
The opening of the file, as written. The whole thing — 103 lines — stays where its author put it; the contents beside it link to each section on GitHub.
3Notch Agent Guide
Purpose
3Notch is a local-first CLI and MCP server for moving project context across boundaries that built-in AI tooling cannot cross — across repos, across AI work surfaces, web chats, and into new projects. V3 is the current shipped surface: packet transfer with artifact bundles, private context seeding, targeted briefs, self-addressed marks, typed packet replies, web-chat stdin intake, relationship indexing, deterministic corpus checks, and .notchpkt pack/unpack for cross-machine transport. Supporting commands: onboard, status, doctor, mcp serve.
Read First
V3 is shipped. Version plans are archived under docs/archived-plans/ as historical record:
docs/archived-plans/v3/3notch-v3-plan.md— current architecture (folder-canonical packets, artifact bundles,nextSteps, pack/unpack).docs/archived-plans/v2/3notch-v2-plan.md— V2 substrate (marks, replies,supersedes, relationships, check, web-chat bridge).docs/archived-plans/v1.1/3notch-v1.1-plan.md— V1.1 hardening.docs/archived-plans/v1/3notch-v1-technical-spec.md— V1 architecture and contract.docs/archived-plans/3notch-project-request.md— original project framing.
Future version plans land under docs/active-plans/<version>/ during development and move to docs/archived-plans/ when shipped.
Long-Run Goal
/goal prompts are an end-of-session deliverable, not a repo artifact. At the close of a planning or build session, when the next logical build is in view, produce a comprehensive paste-ready /goal prompt for that next session and deliver it inline in the chat as a copy-paste Markdown block. Do not write /goal prompts to files in the repo — the user copies them into the next session manually.
Constraints on the prompt:
- Keep it under 4,000 characters.
- Reference the relevant repo docs (the active plan, schemas, services) by path rather than embedding the full plan inline.
- Include only the load-bearing framing the next session needs to act on, not the reasoning that produced the plan.
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 · 103 lines · 1,527 tokens per session scan A 1c6a731b399b
3notch AGENTS.md is an instructions file published in the GitHub repository coldlogicAI/3notch (4 stars, last pushed 21d ago), licensed MIT. It adds 1,527 tokens to every session, about $0.0076 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.
Other instructions, from other repositories
cocoindex AGENTS.md
AGENTS.md instructions for cocoindex-io/cocoindex, covering agents.md, build and test commands, building, testing and code formatting and linting.
apm python.instructions.md
Python development guidelines.
framework AGENTS.md
Instructions for ai-driven-dev/framework, covering agents.md, behavior, communication, action and memory management.
packmind packmind-rest-api-endpoint-design.instructions.md
Instructions for PackmindHub/packmind: Conventions for designing REST API endpoints that are predictable, self-documenting, and aligned with distinct business actions. Favors dedicated action endpoints over generic status updates, and rout... .
Commit Message
Comprehensive Conventional Commits guidelines for openwebui-extensions.
SwarmAI CLAUDE.md
Instructions for xg-gh-25/SwarmAI, covering claude.md, before you push — the local quality gate, commit identity, dev commands and conventions & invariants.