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 skills/robcsaszar/ai-forge/ai-forge-createnpx skills add robcsaszar/ai-forge --skill ai-forge-creategit clone --depth 1 https://github.com/robcsaszar/ai-forgeWrote 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/robcsaszar/ai-forge/ai-forge-create)<a href="https://agentmods.dev/skills/robcsaszar/ai-forge/ai-forge-create"><img src="https://agentmods.dev/badge/skills/robcsaszar/ai-forge/ai-forge-create.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 | $0.00112 | $0.02957 |
| Opus 5 | $0.00056 | $0.01478 |
| Sonnet 5 | $0.00022 | $0.00591 |
| Haiku 4.5 | $0.00011 | $0.00296 |
Grade A, and why
ai-forge-create 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 3d 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 — 200 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AI Forge Create
Write skills and agents that score ≥B on ai-forge-judge out of the box. Every section must earn its tokens.
Phase 0 — Dedup
Before anything else, check whether this already exists. Glob the platform roots listed in ai-forge-audit Phase 1 and compare names and descriptions against the request.
- Strong match (same domain and overlapping triggers) — surface it and stop: "
<name>already covers this. Options — (u)pdate it via ai-forge-update, (c)reate anyway with a narrower scope, or (q)uit." An explicit "create a skill" request plus a strong match is a reason to challenge, not to comply. - Partial match — name it, then continue; the new artifact must state how it differs in its own description.
- No match — continue silently.
Phase 1 — Discovery
Build understanding through a running recap. Ask questions to fill gaps; after each exchange, show the current recap:
## Discovery Recap
**Domain:** [what the skill/agent covers, or "unknown"]
**Artifact type:** [Skill / Agent / Instruction file]
**Platform:** [Claude Code / GitHub Copilot / OpenAI Codex / Google Gemini — infer from file path: .claude/ → Claude Code, .github/ → Copilot, .codex/ → Codex, .gemini/ → Gemini]
**Decisions:** [non-obvious choices the agent must make, or "unknown"]
**Failure modes:** [what breaks without this artifact, or "unknown"]
**Audience:** [fragile/creative output, or "inferred: ..."]
**Size:** [reference files needed, or "inferred: self-contained"]
Then ask: (a)ccept / (r)evise / (q)uit
a— recap is complete; proceed to Phase 2r— user adds or corrects; update recap and loopq— abort
Do not proceed to Phase 2 until the user accepts the recap. Domain, artifact type, and failure modes must be filled before accepting. Platform must also be resolved — it drives frontmatter schema, file location, and invocation constraints for both agents and skills.
Porting an existing artifact to another platform? That's a conversion, not authoring — hand off to
ai-forge-update, which runs the target disposition report and steps the judgment calls through approval. The per-field mappings live inreferences/conversion-guide.md.
What ships with it
10 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- assets/SKILL.template.md 1.6 KB
- references/agent-patterns.md 7.4 KB
- references/agents-taxonomy.md 11 KB
- references/baseline-probe.md 4.0 KB
- references/conversion-guide.md 6.2 KB
- references/drafting-craft.md 3.2 KB
- references/skill-checklist.md 3.4 KB
- references/skill-patterns.md 3.7 KB
- references/skills-taxonomy.md 7.6 KB
- scripts/validate-metadata.cjs 12 KB runs code
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.
- 3d ago First seen · 200 lines · 112 tokens per session scan A 0291950f5490
ai-forge-create is a skill published in the GitHub repository robcsaszar/ai-forge (0 stars, last pushed 4d ago), licensed MIT. It adds 112 tokens to every session and 2,957 once invoked, about $0.0006 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 2026-08-31.
Other skills, from other repositories
iron-system
Meta-skill for reverse loop engineering that transforms an AI agent's approach to any complex task. Combines systematic deconstruction (reverse engineering) and iterative construction (loop engineering) in an EXTRACT→SYNTHESIZE→BUILD→AUDIT cycle. Includes 7 named tactical loops (FORGE, SIEGE, PRISM, RAZOR, MIRROR…
skill-creator
Create or optimize Claude Code skills with best-practice structure, bundled resources, and prompt engineering patterns.
dspy
Build complex AI systems with declarative programming, optimize prompts automatically, create modular RAG systems and agents with DSPy - Stanford NLP's framework for systematic LM programming.
guidance
Control LLM output with regex and grammars, guarantee valid JSON/XML/code generation, enforce structured formats, and build multi-step workflows with Guidance - Microsoft Research's constrained generation framework.
instructor
Extract structured data from LLM responses with Pydantic validation, retry failed extractions automatically, parse complex JSON with type safety, and stream partial results with Instructor - battle-tested structured output library.
outlines
Guarantee valid JSON/XML/code structure during generation, use Pydantic models for type-safe outputs, support local models (Transformers, vLLM), and maximize inference speed with Outlines - dottxt.ai's structured generation library.