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/iusztinpaul/squid/agents-mdgit clone --depth 1 https://github.com/iusztinpaul/squidWrote 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/iusztinpaul/squid/agents-md)<a href="https://agentmods.dev/instructions/iusztinpaul/squid/agents-md"><img src="https://agentmods.dev/badge/instructions/iusztinpaul/squid/agents-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 | $0.03439 | $0.03439 |
| Opus 5 | $0.01720 | $0.01720 |
| Sonnet 5 | $0.00688 | $0.00688 |
| Haiku 4.5 | $0.00344 | $0.00344 |
Grade B, and why
squid AGENTS.md scanned grade B with 1 finding 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 3d 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.
Enumerates other installed skillsmediumAgent snooping
Other skills' SKILL.md files reveal prompts, capabilities and secrets that should be invisible to peers.
- **Test `/squid-scaffold`:** run it in an empty directory and confirm it produces a sensible `AGENTS.md` (with `CLAUDE.md` symlinked to it, plus a `.agents/skills/` dir and a `.claude/skills → .agents/skills` symlink) a How it starts
The opening of the file, as written. The whole thing — 121 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Squid: An Opinionated Software Factory for Claude Code
This repo is a Claude Code plugin. The contract layer is markdown — agents and skills. Nothing else ships; there is no build step, no test suite, no language-specific runtime tooling.
Consumers install it two ways:
/plugin marketplace add iusztinpaul/squid && /plugin install squid@iusztinpaul— global, all Claude Code sessions. The repo is a one-plugin marketplace (.claude-plugin/marketplace.json) namediusztinpaul(the GitHub account) cataloguing thesquidplugin (the repo). The plugin'ssourceis fixed togithub:iusztinpaul/squid, so the install always fetches from GitHub even when the marketplace itself was added from a local path — relative-path plugin sources aren't supported on Claude Code v2.1.x when the marketplace lives on the local filesystem.claude --plugin-dir /path/to/squid— for plugin-development sessions only; loads the plugin from the working tree without registering or installing anything.
Either way they get an opinionated agent team plus a /squid-scaffold flow that bootstraps new repos from a light spec library. The plugin's job is to teach Claude Code how to build software the way this team builds it — nothing gets shipped as scaffolded files; everything gets generated by agents that read these specs.
For users of this plugin: read
README.md. This file is for people editing the plugin itself.
Key Principles You Will Respect All Over Your Work
- Prefer removing instructions over adding them; write docs, code, and rules with the minimum words that achieve the goal.
- Whenever you add a new rule within the memory (such as AGENTS.md), resources or skills, support it with a clear, concise explanation, plus a set of good and bad examples. Good examples: "a 200-token chunk size", "sub-100ms latency". Bad examples: "a powerful architecture", "a robust pipeline".
What's in the repo
.
├── .claude-plugin/
│ ├── plugin.json # Claude Code plugin manifest (name, version, description)
│ └── marketplace.json # one-plugin marketplace catalog so `/plugin install squid@iusztinpaul` works
├── agents/ # five sub-agents
│ ├── product-architect.md
│ ├── software-engineer.md
│ ├── tester.md
│ ├── pr-reviewer.md
│ └── oncall-engineer.md
├── skills/
│ ├── squid-scaffold/ # /squid-scaffold — bootstrap a new repo (create) or audit drift (evaluate)
│ │ ├── SKILL.md
│ │ ├── rules.md # single source of truth for scaffold rules (create + evaluate)
│ │ ├── AGENTS_TEMPLATE.md # template body /squid-scaffold distils into each project's AGENTS.md
│ │ └── specs/ # reference specs read by scaffold
│ ├── squid-plan/ # /squid-plan — feature spec → approved Tasks Plan (+ worktree)
│ ├── squid-implement-task/ # /squid-implement-task — autonomous task loop (SWE↔Tester, commit each)
│ ├── squid-implement-night/ # /squid-implement-night — end-to-end pipeline (thin orchestrator)
│ ├── squid-review/ # /squid-review — push + PA acceptance + PR-Reviewer
│ ├── squid-review-ci/ # /squid-review-ci — On-Call drives CI green
│ ├── squid-grilling/ # /squid-grilling — stress-test a plan (used inside /squid-plan; adapted from mattpocock/skills, MIT)
│ ├── squid-testing-python/ # test-writing conventions
│ ├── squid-self-improve/ # end-of-session corrections capture (human-gated by /squid-implement-night)
│ ├── squid-refactor/ # /squid-refactor — standalone refactor planner (not wired into the pipeline)
│ ├── squid-triage-issue/ # /squid-triage-issue — standalone bug intake (not wired into the pipeline)
│ ├── squid-architecture-review/ # /squid-architecture-review — standalone audit (not wired into the pipeline)
│ ├── squid-clean-docs/ # /squid-clean-docs — standalone prose cleanup: docs, comments, docstrings (not wired into the pipeline)
│ ├── squid-clean-memory/ # /squid-clean-memory — standalone memory-file shrink: compress, dedupe, merge (not wired into the pipeline)
│ ├── squid-clean-harness/ # /squid-clean-harness — standalone .agents shrink: skills + resources, same logic fewer tokens (not wired into the pipeline)
│ └── squid-write-skill/ # /squid-write-skill — skill-authoring reference (vendored from mattpocock/skills, MIT; not wired into the pipeline)
├── LICENSE
├── README.md # user-facing (install + what's inside)
├── AGENTS.md # (this file; CLAUDE.md symlinks to it)
└── .gitignore
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.
- 3d ago First seen · 121 lines · 3,439 tokens per session scan B aa01bf036bcf
squid AGENTS.md is an instructions file published in the GitHub repository iusztinpaul/squid (184 stars, last pushed 1mo ago), licensed Apache-2.0. It adds 3,439 tokens to every session, about $0.0172 per session on Opus 5. A static security scan graded it B with 1 finding (enumerates other installed skills). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-30.
Other instructions, from other repositories
ai-assist-skills AGENTS.md
AGENTS.md instructions for jparkerweb/ai-assist-skills, covering agents.md, project overview, git commit messages, local development and install a single skill from your local working tree.
ai-assist-skills CLAUDE.md
Claude Code instructions for jparkerweb/ai-assist-skills: CRITICAL — MANDATORY FIRST STEP: You MUST read AGENTS.md before responding to ANY user message, including simple questions. Do NOT skip this step regardless of how trivial the request appears. No exceptions.
wizard AGENTS.md
AGENTS.md instructions for redhuntlabs/wizard, covering wizard — for ai agents, what this plugin gives you, your behavioral commitments, where things live and where to look first.
wizard CLAUDE.md
Claude Code instructions for redhuntlabs/wizard, covering wizard — for claude, quick orientation, how to use skills in claude code, how to dispatch the tester and slash commands.
wizard GEMINI.md
Gemini CLI instructions for redhuntlabs/wizard, covering wizard — for gemini, quick orientation, how to use skills in gemini cli, tool name mapping and the 1% rule.
vibe-coding-prompt-template backend.instructions.md
Instructions for KhazP/vibe-coding-prompt-template: Read AGENTS.md, agentdocs/techstack.md, and agentdocs/codepatterns.md.