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.
git clone --depth 1 https://github.com/peva3/anchorWrote 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/rules/peva3/anchor/project-rules)<a href="https://agentmods.dev/rules/peva3/anchor/project-rules"><img src="https://agentmods.dev/badge/rules/peva3/anchor/project-rules/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/rules/peva3/anchor/project-rules"><img src="https://agentmods.dev/badge/rules/peva3/anchor/project-rules.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 | $0.00595 | $0.00595 |
| Opus 5.5 | $0.00238 | $0.00238 |
| Sonnet 5.5 | $0.00119 | $0.00119 |
| Haiku 4.5 | $0.00060 | $0.00060 |
Grade A, and why
project-rules 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 — 76 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Cursor AI Rules
These rules are loaded automatically for this project. See
AGENTS.mdfor full details.
Core Directives
- Read AGENTS.md before any task — All rules are defined there
- Read DEEPDIVE.md — System architecture narrative
- Follow every section — Core principles, commit protocol, testing, linting, etc.
Critical Enforcements
Pre-Commit Checklist
ruff check .
ruff format .
vulture .
mypy .
Do not commit until all pass. (Section 9)
Required Folders
Every project MUST have:
tests/— All tests + random scripts (in.gitignore)docs/— Architecture documentationresearch/— Whitepapers, references
Commit Style
Human-sounding, WHY-focused messages. Example:
api: Added JWT refresh to prevent 401 loops
The token refresh had a race condition where concurrent requests
could trigger multiple refresh calls. Added mutex lock around auth state.
See Section 15 for the full pattern.
PR Standards (Section 33)
- 800 lines max, 500 for complex logic
- Single feature per PR
- Use the PR template (Section 35)
AI Code Quality (Section 34)
Self-check: think first, spot laziness, uncertainty, bloat.
Pre-Commit Hooks (Section 37)
Mandatory. Install once: pre-commit install. Hooks run on every commit.
Decision Ladder (Section 50.1)
Before any implementation: YAGNI → stdlib → native → existing dep → one line → minimum.
Tradeoff Comments (Section 50.2)
For every intentional shortcut, document with # ponytail: <what> / # Upgrade to <what> if <trigger>.
Prohibited
- No PRs/issues/GitHub activity without explicit user approval (Section 2)
- No dead code, unused imports, silent exception handlers (Section 1)
- No commits without running lint/vulture sweep first (Section 9)
- No mutable default arguments, bare excepts, eval/exec (Section 27.2)
- No force push to shared branches (Section 36.2)
- No PRs over 800 lines (Section 33)
- No skipping pre-commit hooks (Section 37)
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 First seen · 76 lines · 595 tokens per session scan A e59334308bd7
project-rules is a cursor rule published in the GitHub repository peva3/anchor (8 stars, last pushed 1mo ago), licensed MIT. It adds 595 tokens to every session, about $0.0024 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 2026-08-31.
Other cursor rules, from other repositories
commit
Commit rules — branch safety, explicit staging, 3-option commit messages, push approval.
windsurfrules
Pare wraps CLI tools in MCP servers returning structured JSON. Always prefer Pare MCP tools over raw CLI.
commit
Quality gate runs FIRST. Do NOT commit if any gate fails. Do NOT ask for permission or skip the gate.
git-conventions
Apply when committing, branching, or opening a pull request. Conventional commit format, branch naming, PR scope.
git-workflow
Git workflow: commits, branches, PRs, history management.
git-workflow
Git commit message conventions, PR workflow, and planning approach.