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/lanes-sh/link/claude-mdgit clone --depth 1 https://github.com/lanes-sh/linkWrote 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/lanes-sh/link/claude-md)<a href="https://agentmods.dev/instructions/lanes-sh/link/claude-md"><img src="https://agentmods.dev/badge/instructions/lanes-sh/link/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 | $0.02998 | $0.02998 |
| Opus 5 | $0.01499 | $0.01499 |
| Sonnet 5 | $0.00600 | $0.00600 |
| Haiku 4.5 | $0.00300 | $0.00300 |
Grade A, and why
link 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.
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.
Working in this repository
CONTRIBUTING.md is the substance — the two commands that must pass, the
list of things that will be pushed back on, and why commit messages carry the reasoning. Read
it. This file covers only what an agent gets wrong that a human reading that file would not.
Work on a branch, in a worktree
Implementation goes on a new branch in its own worktree. Never on develop or main: those
are the checkouts the operator has open, and unreviewed work landing there mixes into the files
they are reading.
Never check develop or main out in a worktree either. Git allows a branch in one place at
a time, so a worktree holding develop locks the primary checkout out of its own integration
branch, and the operator cannot pull, rebase or start the next branch without first working out
which worktree took it. A worktree holds a feature branch and nothing else.
$ git worktree add .worktrees/<name> -b <name>
$ cd .worktrees/<name>
$ bun install
$ bun test # establish the baseline before changing anything
.worktrees/ is already in .gitignore. The bun install is not optional: a fresh worktree
has no node_modules. Getting a green baseline first is what makes a later failure
attributable.
A worktree is deleted when its pull request merges, not left for later. Eleven of them across four repositories, all merged, cost 11.5 GB and made the branch list unreadable.
$ git worktree remove .worktrees/<name>
$ git branch -d <name>
Work lands by pull request into develop, squashed — gh pr merge <n> --squash --delete-branch.
Both develop and main require a passing ci and an approval, so a solo merge adds --admin.
The one pull request that is not squashed is the release; see below.
No Claude session links in pull requests. Do not append a https://claude.ai/code/session_…
URL, or a "Generated with Claude Code" footer, to a pull request body. This repository is public
and the link is noise to everyone reading it. The Co-Authored-By trailer on commits is fine.
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.
- today Changed · +18 lines · +228 tokens per session 4008fc57da89
- yesterday Changed · +7 lines · +168 tokens per session 4ff0d35f38bd
- 5d ago First seen · 173 lines · 2,602 tokens per session scan A 5d2d61287ef2
link CLAUDE.md is an instructions file published in the GitHub repository lanes-sh/link (22 stars, last pushed today), licensed Apache-2.0. It adds 2,998 tokens to every session, about $0.0150 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
ebay-mcp AGENTS.md
Instructions for YosefHayim/ebay-mcp, covering agents.md, what this is, instruction sources, validation commands and module map (src/).
mcp-odoo CLAUDE.md
Claude Code instructions for erpipe-org/mcp-odoo: Guidance for Claude Code (claude.ai/code) working in this repository.
ebay-mcp copilot-instructions.md
Instructions for YosefHayim/ebay-mcp, covering github copilot instructions, core expectations, schema and api patterns, documentation and tests and validation.
teamxray copilot-instructions.md
Copilot instructions for AndreaGriffiths11/teamxray, covering github copilot instructions for mcp team x-ray extension, response style, build, test, and lint commands, validating extension changes and architecture.
teamxray AGENTS.md
AGENTS.md instructions for AndreaGriffiths11/teamxray, covering agents.md, purpose, tagline, tech stack and architecture.
ebay-mcp CLAUDE.md
Instructions for YosefHayim/ebay-mcp, covering claude.md and claude-specific notes.