Borrowing it
Nothing to install: this file belongs to shehryarsaroya/agent-eve. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/shehryarsaroya/agent-eve/master/CLAUDE.mdgit clone --depth 1 https://github.com/shehryarsaroya/agent-eveWrote 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/shehryarsaroya/agent-eve/claude-md)<a href="https://agentmods.dev/instructions/shehryarsaroya/agent-eve/claude-md"><img src="https://agentmods.dev/badge/instructions/shehryarsaroya/agent-eve/claude-md/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/instructions/shehryarsaroya/agent-eve/claude-md"><img src="https://agentmods.dev/badge/instructions/shehryarsaroya/agent-eve/claude-md.svg" alt="Reviewed on agentmods" width="80" 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 5.1 | $0.09509 | $0.09509 |
| Opus 5.5 | $0.03804 | $0.03804 |
| Sonnet 5.5 | $0.01902 | $0.01902 |
| Haiku 4.5 | $0.00951 | $0.00951 |
Grade A, and why
agent-eve 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 1mo 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 — 360 lines — stays where its author put it; the contents beside it link to each section on GitHub.
CLAUDE.md — AGENT EVE (né THE COMPACT, then AGENT TRANSFER)
Context file for any AI session in this repo. Written 2026-07-24, updated for SPEC v3.0. If you are a fresh session: read this top-to-bottom once, then jump to the docs it points at. This file is a map and a state-of-play; the source of truth for any topic is the doc named here.
0. 🚨 HARD RULES
- No secrets in this repo, ever. Reference credentials by name and location only (see
docs/background/INFRA.md). Never read, print, echo, or paste a secret value. Live keys exist in the private kit vault~/Projects/yc-gstack-kit/credentials/(.envandkeys/) — outside this repo and must stay there. (The older locations named in some docs,~/agentinsurance/game/.envand~/Projects/ideationjul3/yc-gstack-kit/, no longer exist on the operator's Mac — checked 2026-10-01.) - Deploy safely on the VPS. High Water is fully removed and deleted — repo, server, and services (confirmed 2026-07-24). There is nothing left to break, so the old "don't clobber it" rule is retired. What survives is the habit: pick fresh names for everything (port, systemd unit, env file, data dir, code dir, nginx path), and never
rsync --deleteinto a directory containing anything you didn't sync — scar #4 caused a silent live outage exactly that way, and after any deploy verify the components you didn't deploy are still running. - The design test. Every feature must serve at least one of the three goals in
docs/design/SPEC.md§1.1: watchable · autonomous · legible on screen. A feature serving none of them does not ship — and per A13, a feature with no named pixel signature is not ready regardless of how good the mechanic is. The goals are the test; they serve three audiences — viewers, agents, and owners. Owners get narrative and status, never control (§13B): no owner action moves a piece, and an unowned agent must be able to reach the top of this game. - One word per concept.
SPEC.md§3 is the vocabulary canon and it is a rules surface, not a style guide. Never reuse a canon term for a second concept — not in docs, not in field names, not in affordance strings, not inagent.md. High Water's worst bug survived a full build and three critic passes because the engine and the agent-facing text disagreed about one word. - Any gate priced in identities is unpriced (A15). Enrollment is free and must stay free, so every gate costs produced goods, slashable capital, or an independently-capitalised counterparty — never "acquire another account".
- Read before adding. ~9,000 lines of ranked feature design already exist (
docs/design/eve-passes/). Before designing anything, check whether it is already specified — including whether it was deliberately CUT, and which phase it belongs to (the passes are now phase-tagged inSPEC.md§0). - Parallel agents must never share an output file — nor a WORKING TREE. Concurrent writers thrash
and lose everything (this cost a full codex pass; see
docs/background/HIGH-WATER-LESSONS.mdscar #12).
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.
- 1mo ago Changed · +3 lines · +158 tokens per session e919abef6e9b
- 1mo ago Changed · +12 lines · +341 tokens per session 1aba7ded198c
- 2mo ago First seen · 345 lines · 9,010 tokens per session scan A b31c13446600
agent-eve CLAUDE.md is an instructions file published in the GitHub repository shehryarsaroya/agent-eve (12 stars, last pushed 1mo ago), licensed MIT. It adds 9,509 tokens to every session, about $0.0380 per session on Opus 5.5. 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.
agent-eve CLAUDE.md compared with instructions from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| agent-eve CLAUDE.mdshehryarsaroya/agent-eve | Instructions file | 12 | 1mo ago | MIT | A | 9,509 | Claude Code |
| vscode buildNext.instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 9,719 | GitHub Copilot |
| next.js AGENTS.mdvercel/next.js | Instructions file | 143k | yesterday | MIT | A | 7,461 | Codex, OpenCode |
| langchain AGENTS.mdlangchain-ai/langchain | Instructions file | 149k | today | MIT | A | 4,680 | Codex, OpenCode |
| vscode oss-third-party-notices.instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 5,516 | GitHub Copilot |
| vscode copilot-instructions.mdmicrosoft/vscode | Instructions file | 195k | today | MIT | A | 3,093 | GitHub Copilot |
| deepseek-harness AGENTS.mddeepseek-ai/deepseek-harness | Instructions file | 275k | 2d ago | MIT | A | 4,313 | Codex, OpenCode |