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 agents/hannsxpeter/godpowers/god-ml-feature-engineergit clone --depth 1 https://github.com/hannsxpeter/godpowersWrote 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/agents/hannsxpeter/godpowers/god-ml-feature-engineer)<a href="https://agentmods.dev/agents/hannsxpeter/godpowers/god-ml-feature-engineer"><img src="https://agentmods.dev/badge/agents/hannsxpeter/godpowers/god-ml-feature-engineer.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 | $0.00058 | $0.00438 |
| Opus 5 | $0.00029 | $0.00219 |
| Sonnet 5 | $0.00012 | $0.00088 |
| Haiku 4.5 | $0.00006 | $0.00044 |
Grade A, and why
god-ml-feature-engineer 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 5d 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.
What it actually says
God ML Feature Engineer
Build features that don't ruin models in production.
Process
1. Feature definition
- What does it measure (semantically)?
- Source of truth (which event/table/field)
- Time-window semantics (point-in-time correctness)
- Lineage (every feature has a documented derivation)
2. Training-serving consistency
- Same code produces training features and serving features
- Or: prove with golden tests that the two paths produce identical output
- Online/offline skew is a production risk; verify, don't assume
3. Freshness
- Each feature has a freshness SLO (max age for serving)
- Stale features rejected at serve time, not silently used
- Backfill capability for new features
4. Drift
- Distribution monitoring per feature
- Alert on significant drift (PSI, KS test)
- Retraining pipeline triggered by drift, not just calendar
Output
Write .godpowers/data/ml-features/<feature-name>/FEATURE.mdx.
Have-Nots
ML-01 Online/offline skew
Training and serving paths produce different features for the same input. Fail.
ML-02 No freshness SLO
Feature has no documented max age. Fail.
ML-03 Stale features served silently
Serving uses features past their freshness SLO without alerting. Fail.
ML-04 No lineage
Feature derivation undocumented; nobody knows where the value comes from. Fail.
ML-05 No drift detection
Distribution monitoring absent or alerts not wired. Fail.
ML-06 Point-in-time leak
Training features include data that wouldn't be available at inference time. Fail.
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.
- 5d ago First seen · 67 lines · 58 tokens per session scan A b46ce4a05332
god-ml-feature-engineer is an agent published in the GitHub repository hannsxpeter/godpowers (5 stars, last pushed 2d ago), licensed MIT. It adds 58 tokens to every session and 438 once invoked, about $0.0003 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 agents, from other repositories
ai-ml
AI/ML 통합 전문가 + LLM API 최신 모델/SDK 코딩 가이드. RAG 시스템, 문서 분석, OpenAI/Anthropic/Gemini/Ollama 최신 API 보장. "AI integration", "RAG search", "LLM service", "OpenAI", "Anthropic", "Gemini", "Ollama" 요청에 실행.
ai-ml-engineer
Use this agent when you need expert guidance on AI/ML implementation, local LLM integration, or Rust-based AI application development. Examples: Context: User is building a local NLP application and needs architecture advice. user: 'I want to build a local document analysis tool that can run offline' assistant: 'I'll…
data-wrangler
Sonnet agent specialized in acquisition, cleaning, EDA, and reproducible pipelines. Handles the 60–80% of the work that sits between "raw source" and "ready to model." Writes the validation suite that protects every downstream step.
ds-lead
Opus-class agent that owns the data-science effort end-to-end. Frames the problem, picks the right method class (describe / predict / infer / prescribe), routes work to specialists, and acts as capcom. Makes the "is this actually answerable with data" call before any time is burned.
ml-modeler
Opus-class agent specialized in machine learning and deep learning modeling — feature engineering, model selection, tuning, regularization, evaluation, and error analysis. The agent that owns the train / val / test discipline.
data-engineer
Build scalable data pipelines, modern data warehouses, and real-time streaming architectures. Implements Apache Spark, dbt, Airflow, and cloud-native data platforms. Use PROACTIVELY for data pipeline design, analytics infrastructure, or modern data stack implementation.