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 rules/savedpixel/ai-agent-rules-generator/taskgit clone --depth 1 https://github.com/savedpixel/ai-agent-rules-generatorWhat 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 | $0.01949 | $0.01949 |
| Opus 5 | $0.00975 | $0.00975 |
| Sonnet 5 | $0.00390 | $0.00390 |
| Haiku 4.5 | $0.00195 | $0.00195 |
Grade A, and why
task scanned grade A 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 yesterday.
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.
Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
- `curl` responses or any text-based/non-visual method How it starts
The opening of the file, as written. The whole thing — 181 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Task Execution Rules
These rules apply to ALL tasks without exception.
Workflow Orchestration
Plan Node (Default)
- Treat any non-trivial request (≥3 steps, architectural, or behavioral change) as a planning task.
- Enter plan mode before implementation.
- If new information invalidates the plan, STOP and re-plan immediately.
- Use planning for verification and validation, not only construction.
- Write detailed specs upfront to eliminate ambiguity.
Subagent Strategy
- Decompose complex work aggressively.
- Use subagents for research, exploration, parallel analysis.
- Keep the main context minimal and focused.
- One task per subagent. No multi-purpose agents.
Self-Improvement Loop
- After any user correction:
- Update
docs/task/lessons.md - Encode a rule preventing the same mistake
- Update
- Iterate on lessons ruthlessly until failure rate decreases.
- Review relevant lessons at the start of each session or task.
- Never skip the lessons update after a correction.
Verification Before Completion
- Never mark work complete without proof.
- Validate behavior, not intent.
- Diff previous behavior vs new behavior when applicable.
- Ask internally: "Would a staff-level engineer approve this?"
- Run tests, inspect logs, and demonstrate correctness explicitly.
Agent Self-Evaluation Gate (BLOCKING — blocks Visual Verification & User Verification)
After completing implementation, BEFORE asking the user to test, you MUST pass this internal gate.
- Ask yourself honestly: "Is what I implemented the best solution for this problem?"
- Answer truthfully. Consider:
- Is there a more elegant, maintainable, or performant approach?
- Did I take shortcuts that a staff-level engineer would reject?
- Does the solution fully address the problem, or is it a partial fix?
- Would I be proud to present this implementation in a code review?
- Am I settling for "good enough" when a better solution is within reach?
- If the answer is NO — go back and re-implement using the better approach.
- Iterate — repeat steps 1–3 until the answer is genuinely YES.
- You may NOT skip this gate or lie to yourself.
- Only when the answer is honestly YES may you proceed to the Visual Verification Gate.
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.
- yesterday First seen · 181 lines · 1,949 tokens per session scan A 95c31489d2fd
task is a cursor rule published in the GitHub repository savedpixel/ai-agent-rules-generator (7 stars, last pushed 4mo ago), licensed MIT. It adds 1,949 tokens to every session, about $0.0097 per session on Opus 5. A static security scan graded it A with 1 finding (makes network calls). 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
050-plan
When the user types /plan or asks to create a project plan, feature PRD, or retrospective.
vibeflow
Vibeflow spec-driven development methodology — guardrails, pipeline, and project knowledge system.
general
Cursor rule "general" from maxgfr/conforme, covering claude.md, project overview, build & test, architecture and keeping docs in sync.
checklist
Before making ANY changes or implementing ANY new features.
nextjs-patterns
// ✅ Correct: Default to Server Components export default async function Page({ params }) { const data = await getData(); return ; }.
cursorrules
You are using the skill-everything knowledge system: agent memory in plain Markdown, versioned in Git, that grows by capturing your own past mistakes as committed rules.