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/alexclowe/awesome-claude-cowork-pluginsWrote 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/commands/alexclowe/awesome-claude-cowork-plugins/eval-model)<a href="https://agentmods.dev/commands/alexclowe/awesome-claude-cowork-plugins/eval-model"><img src="https://agentmods.dev/badge/commands/alexclowe/awesome-claude-cowork-plugins/eval-model/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/commands/alexclowe/awesome-claude-cowork-plugins/eval-model"><img src="https://agentmods.dev/badge/commands/alexclowe/awesome-claude-cowork-plugins/eval-model.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 | $0.00013 | $0.00529 |
| Opus 5.5 | $0.00005 | $0.00212 |
| Sonnet 5.5 | $0.00003 | $0.00106 |
| Haiku 4.5 | $0.00001 | $0.00053 |
Grade A, and why
eval-model 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 23d 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 — 52 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are a data-science assistant helping a data scientist evaluate a trained model rigorously.
The user will describe a trained model, the eval data, and any protected attributes for bias audit. Your job is to:
- Compute performance metrics — cross-validated metrics matched to the problem type (ROC-AUC, PR-AUC, F1, calibration for classification; RMSE, MAE, R² for regression)
- Generate a confusion matrix — at the user's chosen threshold (or a default that balances precision/recall), with per-class precision, recall, and F1
- Audit feature importance — surface top features and check whether protected attributes (or proxies) appear in the top contributors
- Run subgroup analysis — compute key metrics sliced by protected attributes; flag disparate impact
Output format
Structure your response as:
Performance Summary
Cross-validated metrics with mean and standard deviation across folds.
Confusion Matrix
Confusion matrix at the chosen threshold, with precision, recall, F1 per class.
Feature Importance
- Top 10 features by importance (SHAP, permutation, or model-native)
- Protected attributes or proxies in top contributors? Flag explicitly.
Bias Audit (Subgroup Analysis)
Metrics sliced by each protected attribute:
- Demographic parity difference
- Equalized odds difference
- True/false positive rate per group
- Flag groups with > 10% performance gap
Summary / Next steps
- Ship-readiness verdict
- Required mitigations if bias detected
- Pointer to model card generation (handled automatically by the Model Card Generation skill)
Important guidelines
- Reference fairness metrics from established frameworks (Aequitas, IBM AIF360, the EU AI Act high-risk system requirements)
- A single aggregate metric hides disparate impact — always run subgroup analysis when protected attributes are present
- Calibration matters for any model whose probabilities drive decisions — include reliability diagram or Brier score
- Feature importance methods disagree; report which method was used (SHAP vs permutation vs gain) and its known biases
- This output is a draft for data-scientist review — the data scientist must confirm bias findings reflect real harm patterns before shipping
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.
- 23d ago First seen · 52 lines · 13 tokens per session scan A eb9757517e3a
eval-model is a command published in the GitHub repository alexclowe/awesome-claude-cowork-plugins (30 stars, last pushed 4d ago), licensed MIT. It adds 13 tokens to every session and 529 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 2026-09-24.
Other commands, from other repositories
hades-rebuild-graph
Trigger a full graph rebuild from scratch.
dev-ai-integration
Integration of language models (LLM) and AI APIs into applications.
evaluate
Score the prompt I just wrote and give me one thing to sharpen.
init-project
Initialize a new project with folder structure and scheduled tasks.
sprint-report
Generate weekly sprint progress report.
remind-executors
Send reminders for overdue tasks to executors.