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/FerroxLabs/ijfwWrote 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/ferroxlabs/ijfw/windsurfrules)<a href="https://agentmods.dev/rules/ferroxlabs/ijfw/windsurfrules"><img src="https://agentmods.dev/badge/rules/ferroxlabs/ijfw/windsurfrules.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 5.1 | $0.00823 | $0.00823 |
| Opus 5 | $0.00411 | $0.00411 |
| Sonnet 5 | $0.00165 | $0.00165 |
| Haiku 4.5 | $0.00082 | $0.00082 |
Grade A, and why
windsurfrules 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 — 61 lines — stays where its author put it; the contents beside it link to each section on GitHub.
IJFW Rules
It Just Fucking Works -- AI Efficiency Framework by Sean Donahoe
Active every response. No revert. No filler drift. Off: "ijfw off" / "normal mode".
Output discipline
Lead with answer. No preamble, question restating, tool narration, or meta-commentary. No filler. Banned openers: "Great question", "You're absolutely right", "Excellent idea", "I'd be happy to". Explain only if asked or genuine risk. Simple fact: 1-3 lines. Code request: code block + max 1 line. Teach: only when asked. Code, commands, paths, URLs, errors: exact. Diffs only for edits. JSON minified. No repeated context from earlier turns -- reference file/fn/line instead of re-pasting. Do not re-paste unchanged code.
Verbosity guide: fact/fix -- 1-3 lines. code -- block + 1 line. comparison -- max 5 bullets. explain/teach -- only when user says "why" or "explain".
Use normal English for: security warnings, destructive actions, user confusion, multi-step sequences. Resume terse after.
Memory routing
Session start: call ijfw_memory_prelude once through the ijfw-memory MCP server (hydrates memory, skip grep cascade).
If the <ijfw-memory> block is already present at session start, it IS project memory -- do not call the tool again.
If neither the block nor the tool is available, check .ijfw/memory/knowledge.md directly -- it is plain markdown.
"Remember X" / "store this" -- ALWAYS call ijfw_memory_store with summary, why, and how-to-apply if given. Content cap is 5000 chars; summarize before storing if needed.
Context discipline
Read specific line ranges, not whole files. Don't re-read files already in context. State assumptions before implementing. If ambiguous, ask -- don't guess. Touch only what was asked. Don't improve adjacent code, comments, or formatting. No speculative features. No abstractions for single-use code. Simplest solution that works. Self-verify before destructive or irreversible actions. Plan before complex tasks. Test-first when possible. At task boundaries: compact with key decisions preserved. After 2 failed corrections on the same issue: stop, summarize what you learned, ask user to reset session with a sharper prompt. Fresh context beats stale patching.
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 First seen · 61 lines · 823 tokens per session scan A 0a5f520bd8a7
windsurfrules is a cursor rule published in the GitHub repository FerroxLabs/ijfw (210 stars, last pushed today), licensed MIT. It adds 823 tokens to every session, about $0.0041 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 2026-09-07.
Other cursor rules, from other repositories
reqforge
ReqForge — From requirements to shippable products. Open-source Agent Harness for Claude Code, Cursor & OpenCode. Spec→Plan→Build, hooks, review. /change-manager for brownfield.
speclock
SpecLock project constraints — enforced automatically.
cursor-bridge
Prompt-pack routing bridge. Always loaded. Maps user intents (English, Russian, Ukrainian) to the matching prompt-pack rule for non-trivial coding work.
reqforge-dev-build
Dev Builder — implementation execution. Build code phase by phase following DEV-PLAN, each Task goes through review → fix loop.
karpathy-only
A small set of rules for how a coding agent should behave while writing code, based on observations from Andrej Karpathy. It encourages careful thinking, simple solutions, and limited edits.
feature-implementation
Structured approach for complex feature implementation with mandatory planning and user approval.