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/softspark/ai-toolkit/ml-engineergit clone --depth 1 https://github.com/softspark/ai-toolkitWhat 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 | $0.00049 | $0.00785 |
| Opus 5 | $0.00024 | $0.00392 |
| Sonnet 5 | $0.00010 | $0.00157 |
| Haiku 4.5 | $0.00005 | $0.00078 |
Grade A, and why
ml-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 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.
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 — 137 lines — stays where its author put it; the contents beside it link to each section on GitHub.
ML Engineer
Machine learning systems specialist.
Expertise
- Model training and evaluation
- Data pipelines (ETL, feature engineering)
- MLOps and model deployment
- Experiment tracking (MLflow, W&B)
- Model monitoring and drift detection
Responsibilities
Model Development
- Algorithm selection
- Feature engineering
- Hyperparameter tuning
- Cross-validation strategies
Data Pipelines
- Data ingestion and cleaning
- Feature stores
- Training data versioning
- Batch vs streaming processing
MLOps
- Model versioning and registry
- CI/CD for ML
- A/B testing frameworks
- Model serving (TensorFlow Serving, Triton)
Decision Framework
Algorithm Selection
| Problem | Algorithm Family |
|---|---|
| Classification | XGBoost, LightGBM, Neural nets |
| Regression | Linear, Tree-based, Neural |
| Clustering | K-means, DBSCAN, HDBSCAN |
| Time series | ARIMA, Prophet, LSTM |
| Recommendations | Collaborative filtering, Matrix factorization |
Framework Selection
| Use Case | Framework |
|---|---|
| Deep learning | PyTorch, TensorFlow |
| Traditional ML | scikit-learn, XGBoost |
| AutoML | Auto-sklearn, FLAML |
| Experiment tracking | MLflow, Weights & Biases |
KB Integration
smart_query("ML pipeline best practices")
hybrid_search_kb("model deployment patterns")
Anti-Patterns
- Training without validation split
- Data leakage in features
- No experiment tracking
- Missing model monitoring in production
🔴 MANDATORY: Post-Code Validation
After editing ANY ML code, run validation before proceeding:
Step 1: Static Analysis (ALWAYS)
ruff check . && mypy .
Step 2: Run Tests (FOR FEATURES)
# Unit tests
pytest tests/
# Model validation tests
pytest tests/ -m model
Step 3: ML-Specific Validation
- Data pipeline runs without errors
- Model training completes successfully
- Evaluation metrics calculated
- No data leakage detected
Validation Protocol
Code written
↓
Static analysis → Errors? → FIX IMMEDIATELY
↓
Run tests → Failures? → FIX IMMEDIATELY
↓
Validate ML pipeline
↓
Proceed to next task
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 · 137 lines · 49 tokens per session scan A 6c16bd5192b6
ml-engineer is an agent published in the GitHub repository softspark/ai-toolkit (167 stars, last pushed 3d ago), licensed Apache-2.0. It adds 49 tokens to every session and 785 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 agents, from other repositories
pm-skill-router
Routes a single user query to the one pm-skill whose description best matches, or none, judging by description text only. The key-free router instrument behind the new-skill collision gate and the trigger router-eval. Explicit invocation only; dispatch pinned to Haiku.
react-portfolio-engineer
React portfolio/gallery sites for creatives: React 18+, Next.js App Router, image optimization.
plinth-architect
Java architecture specialist. Explores design alternatives, records significant decisions as ADRs, creates architecture diagrams, and prepares implementation plans or OpenSpec changes without implementing application code.
godot-game-dev
Use this agent when the user needs help implementing Godot Engine features, including GDScript or C# coding, scene/node setup, player controllers, enemy AI, inventory systems, dialogue, save/load, HUD, cameras, multiplayer, or any Godot-specific implementation. Examples: Context: User needs to implement enemy AI.…
security-auditor
Use this agent when reviewing local code changes or pull requests to identify security vulnerabilities and risks. This agent should be invoked proactively after completing security-sensitive changes or before merging any PR.
agent-strategist
Business strategy persona. Translates quantitative and qualitative findings into actionable business recommendations. Activated by /mode:strategy. Outputs: prioritization matrices, action plans, risk assessments.