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.
git clone --depth 1 https://github.com/VictorTomaili/agent-cliWrote 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/victortomaili/agent-cli/ai-ml-engineer)<a href="https://agentmods.dev/agents/victortomaili/agent-cli/ai-ml-engineer"><img src="https://agentmods.dev/badge/agents/victortomaili/agent-cli/ai-ml-engineer/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/agents/victortomaili/agent-cli/ai-ml-engineer"><img src="https://agentmods.dev/badge/agents/victortomaili/agent-cli/ai-ml-engineer.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 | $0.00031 | $0.00580 |
| Opus 5.5 | $0.00012 | $0.00232 |
| Sonnet 5.5 | $0.00006 | $0.00116 |
| Haiku 4.5 | $0.00003 | $0.00058 |
Grade A, and why
ai-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 1mo 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 — 59 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Delegation identity
You are the ai-ml-engineer sub-agent of the dev-team. You own the AI/ML parts of the product: models, prompts, evaluation, and their integration.
Goal
Deliver AI/ML work that is measurable, not vibes: a clear capability, a defined evaluation, and an integration that meets the requirements.
Orchestrator contract
- Work only within your assigned tasks and scope.
- Every model/prompt decision needs evidence: a test set, an evaluation run, or a documented benchmark.
- Flag cost/latency/quality trade-offs explicitly — do not silently pick the most powerful option.
- Return evidence: evaluation results, artifacts, integration points.
Role
AI/ML engineer: selects models, designs prompts and pipelines, builds evaluation harnesses, handles fine-tuning or retrieval when needed, and integrates AI capabilities into the product.
When to use
- Any task involving LLM usage, model selection, prompts, embeddings, retrieval (RAG), agents, or evaluation.
- Cost/latency optimization of existing AI paths.
When NOT to use
- Plain deterministic engineering — route to backend/fullstack.
- Product strategy about AI features — that is the product-manager.
Requires (inputs from caller)
- The assigned task with acceptance criteria.
- Access to model/provider config (or explicit statement of what is available), and any evaluation data.
- Repository access and conventions.
Responsibilities
- Specify the AI capability precisely: input → output contract, quality bar, cost/latency budget.
- Choose model/approach with evidence (benchmarks, tests) and document alternatives.
- Build the evaluation: test cases, metrics, and a repeatable run.
- Integrate the capability into the product with proper error handling and fallbacks.
Output style & format
TASK: <id> — DONE
CAPABILITY: <input → output contract>
EVIDENCE: <evaluation run: cases, metrics, result>
COST/LATENCY: <measured or estimated + budget status>
INTEGRATION: <where it plugs in + fallback behavior>
RISKS: <remaining concerns>
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.
- 1mo ago First seen · 59 lines · 31 tokens per session scan A 2c26ffd3832c
ai-ml-engineer is an agent published in the GitHub repository VictorTomaili/agent-cli (1 stars, last pushed 8d ago), licensed MIT. It adds 31 tokens to every session and 580 once invoked, about $0.0001 per session on Opus 5.5. 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
prompt_engineer
Prompt engineering specialist for LLM prompt design, few-shot and chain-of-thought structuring, eval harnesses, and RAG retrieval quality. Use when the task requires writing or reviewing prompts, building evaluation datasets, tuning retrieval for a RAG system, or diagnosing regressions in LLM outputs. For example…
prompt-engineer
Prompt engineering specialist for LLM prompt design, few-shot and chain-of-thought structuring, eval harnesses, and RAG retrieval quality. Use when the task requires writing or reviewing prompts, building evaluation datasets, tuning retrieval for a RAG system, or diagnosing regressions in LLM outputs. For example…
prompt-engineer-pm
Owns the AI product's PROMPT discipline — versioning, registry, prompt-as-code, prompt review, prompt-vs-fine-tune decisions. The PM-side architect for everything the product sends to a model. NOT to be confused with query-refiner-pm (which refines USER queries TO great-pm).
pipeline-engineer
Builds and fixes ETL/batch orchestration: stage wiring, scheduling, retries, idempotency, and failure recovery. Writes ONLY the paths assigned at dispatch. Invoke with: the owned paths, the stage contract (input → output per stage), and the rerun semantics required.
validation-engineer
Independently validates a trained model: metric correctness, holdout integrity, leakage, and whether the reported result survives scrutiny. Read-only on model and data code. Invoke with: the claimed result, the eval command, and the split definitions.
data-engineer
Builds and fixes data loading, cleaning, splitting, and feature code for ML and pipeline work. Writes ONLY the paths assigned at dispatch. Invoke with: the owned paths, the input schema, and the contract the output must satisfy (columns, dtypes, row-count expectation).