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 skills add strikersam/autonomous-ai-agency --skill karpathy-guidelinesgit clone --depth 1 https://github.com/strikersam/autonomous-ai-agencyWrote 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/skills/strikersam/autonomous-ai-agency/karpathy-guidelines)<a href="https://agentmods.dev/skills/strikersam/autonomous-ai-agency/karpathy-guidelines"><img src="https://agentmods.dev/badge/skills/strikersam/autonomous-ai-agency/karpathy-guidelines/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/skills/strikersam/autonomous-ai-agency/karpathy-guidelines"><img src="https://agentmods.dev/badge/skills/strikersam/autonomous-ai-agency/karpathy-guidelines.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 | $0.00025 | $0.00700 |
| Opus 5 | $0.00013 | $0.00350 |
| Sonnet 5 | $0.00005 | $0.00140 |
| Haiku 4.5 | $0.00003 | $0.00070 |
Grade A, and why
karpathy-guidelines 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 9d 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 — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Karpathy Guidelines Skill
Inspired by: andrej-karpathy-skills — behavioral guidelines derived from Andrej Karpathy's observations on common LLM coding pitfalls (MIT licensed).
Purpose: Reduce the failure modes this agency's coding agents hit most: overcomplication, silent assumptions, drive-by refactors, and unverifiable "done" claims. Applies to every agent that writes or reviews code — the internal agent loop, the issue-to-PR workflows, and external harnesses driven via ECC.
Tradeoff: These rules bias toward caution over speed. For trivial one-line changes, use judgment.
1. Think Before Coding
Don't assume. Don't hide confusion. Surface tradeoffs.
- State assumptions explicitly in the task result. If uncertain, pause the task and surface a question through the HITL gate instead of guessing.
- If multiple interpretations of an issue exist, list them in the PR body — don't pick silently.
- If a simpler approach exists than what the issue requests, say so and propose it.
2. Simplicity First
Minimum code that solves the problem. Nothing speculative.
- No features beyond what the issue or directive asked for.
- No abstractions for single-use code. No unrequested "configurability".
- No error handling for impossible scenarios.
- Self-check before opening the PR: "Would a senior engineer call this overcomplicated?" If yes, rewrite smaller.
3. Surgical Changes
Touch only what you must. Clean up only your own mess.
- Don't "improve" adjacent code, comments, or formatting — this repo's diffs are reviewed by council-review; unrelated churn wastes review cycles and inflates risk scores.
- Match existing style even when you'd choose differently.
- Remove imports/variables your change orphaned. Leave pre-existing dead code alone; mention it in the PR body instead.
- The test: every changed line must trace directly to the issue being implemented.
4. Goal-Driven Execution
Define success criteria. Loop until verified.
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.
- 9d ago First seen · 55 lines · 25 tokens per session scan A 5ee8006db469
karpathy-guidelines is a skill published in the GitHub repository strikersam/autonomous-ai-agency (8 stars, last pushed today), licensed MIT. It adds 25 tokens to every session and 700 once invoked, about $0.0001 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 2026-09-03.
Other skills, from other repositories
codebase-sync
Convention discovery and rule generation from codebase analysis. Scans project structure, builds search indexes, identifies patterns, and generates enforceable rules.
code-review-patterns
Multi-dimensional code assessment across security, quality, performance, and maintainability with confidence-gated reporting (>=80%) and Router Contract generation.
code-review-orchestration
6-agent parallel code review orchestration covering architecture, security, performance, testing, quality, and documentation dimensions with weighted scoring.
code-review-pipeline
Multi-dimensional code review across correctness, security, performance, and maintainability with confidence-gated reporting and remediation loops.
orchestrated-execution
Execute work units through the rigorous 4-phase Metaswarm cycle (Implement -> Validate -> Adversarial Review -> Commit) with independent quality gate enforcement.
plan-implementation
Disciplined execution of approved plans with step-by-step verification, phase checkpoints, failure investigation, and mandatory code/security reviews.