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 zorost/AI-Engineering-Lab --skill spec-first-ai-featuregit clone --depth 1 https://github.com/zorost/AI-Engineering-LabWrote 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/zorost/ai-engineering-lab/spec-first-ai-feature)<a href="https://agentmods.dev/skills/zorost/ai-engineering-lab/spec-first-ai-feature"><img src="https://agentmods.dev/badge/skills/zorost/ai-engineering-lab/spec-first-ai-feature/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/zorost/ai-engineering-lab/spec-first-ai-feature"><img src="https://agentmods.dev/badge/skills/zorost/ai-engineering-lab/spec-first-ai-feature.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00047 | $0.01064 |
| Opus 5 | $0.00023 | $0.00532 |
| Sonnet 5 | $0.00009 | $0.00213 |
| Haiku 4.5 | $0.00005 | $0.00106 |
Grade A, and why
spec-first-ai-feature 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 10d 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 — 95 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Spec-First AI Feature
1 · Purpose
Force the four decisions that make an AI feature buildable, who it serves, what "correct" means, how it is measured, and what tradeoffs are refused, onto one page before any implementation begins.
2 · When to use
- Starting any AI feature: extraction, RAG bot, agent, fine-tune, integration.
- When a build has drifted for more than a session without a written target.
Do not use for exploratory notebooks whose entire purpose is learning a library. Label those "spike" and timebox them instead.
3 · Inputs
- The feature request, verbatim.
- Access to the requester (or their proxy) for the two questions in step 3.
- One sample of real input data, if any exists.
4 · Procedure
- Create
SPEC.mdin the project root. Use the five headings below, no others. - User & job. Write who uses the output and what decision or action it feeds, in two sentences. If you cannot name the user, STOP and ask.
- Ask the requester two questions and record the answers verbatim:
- "Show me three real inputs and the outputs you would accept."
- "What output would make you reject this on sight?"
- Golden set. Name the file that will hold the accepted input→output pairs. Target 20 cases minimum. If three real cases cannot be obtained, STOP: the feature is not spec-able yet.
- Metric & gate. Write one measurable definition of correctness (e.g., per-field exact-match accuracy) and the pass value (e.g., ≥ 0.90). One metric, one number. Secondary observations (latency, cost) go in a note, not the gate.
- Refused tradeoffs. List what this feature will not sacrifice: e.g. "never invent a missing field", "no PII leaves the VPC", "p95 latency under 4 s". Write at least one. A spec with no refused tradeoff is a wish, not a spec.
- Blast radius. Write what the feature may write to or send, and what requires human approval (deletes, external sends, spend above $X).
- Show the spec to the requester. Record their "yes" or their edits. A spec nobody approved is a draft.
- Only then begin build work. Point every later dispute at this file; update the file when the answer changes.
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.
- 10d ago First seen · 95 lines · 47 tokens per session scan A 0249bf135967
spec-first-ai-feature is a skill published in the GitHub repository zorost/AI-Engineering-Lab (302 stars, last pushed 23d ago), licensed MIT. It adds 47 tokens to every session and 1,064 once invoked, about $0.0002 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 2026-08-30.
Other skills, from other repositories
fine-tuning-os
Use this skill when conducting a professional LLM fine-tuning engagement in Zero-Data mode via the MCP server fine-tuning-os. Triggers on: prestation de fine-tuning LLM, livraison de modèle affiné, pipeline Zero-Data, orchestration des 64 outils fine-tuning-os, cycle de vie entraînement (préparation / données…
ai-engineering-toolkit
6 production-ready AI engineering workflows: prompt evaluation (8-dimension scoring), context budget planning, RAG pipeline design, agent security audit (65-point checklist), eval harness building, and product sense coaching.
llm-ops
LLM Operations -- RAG, embeddings, vector databases, fine-tuning, prompt engineering avancado, custos de LLM, evals de qualidade e arquiteturas de IA para producao.
llama-factory
Expert guidance for fine-tuning LLMs with LLaMA-Factory - WebUI no-code, 100+ models, 2/3/4/5/6/8-bit QLoRA, multimodal support.
tinker-training-cost
Calculates training costs for Tinker fine-tuning jobs. Use when estimating costs for Tinker LLM training, counting tokens in datasets, or comparing Tinker model training prices. Tokenizes datasets using the correct model tokenizer and provides accurate cost estimates.
transformers
This skill should be used when working with pre-trained transformer models for natural language processing, computer vision, audio, or multimodal tasks. Use for text generation, classification, question answering, translation, summarization, image classification, object detection, speech recognition, and fine-tuning…