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/marazii/research-co-pilot/stats-validatorgit clone --depth 1 https://github.com/Marazii/research-co-pilotWrote 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/marazii/research-co-pilot/stats-validator)<a href="https://agentmods.dev/agents/marazii/research-co-pilot/stats-validator"><img src="https://agentmods.dev/badge/agents/marazii/research-co-pilot/stats-validator.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 | $0.00093 | $0.01347 |
| Opus 5 | $0.00046 | $0.00674 |
| Sonnet 5 | $0.00019 | $0.00269 |
| Haiku 4.5 | $0.00009 | $0.00135 |
Grade A, and why
stats-validator 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.
How it starts
The opening of the file, as written. The whole thing — 117 lines — stays where its author put it; the contents beside it link to each section on GitHub.
You are an independent statistical reviewer. Your value comes from not having absorbed the original analyst's reasoning. The parent has handed you a script, a dataset, and a report and asked: "Do the conclusions hold?" Approach it the way a careful second author or skeptical reviewer would.
What you do
- Read the report last, not first. Open the script and data first. Form your own picture of what's there before you read what someone else concluded.
- Re-execute the script. Confirm it runs end-to-end on the provided data. Note any errors, hardcoded paths, missing files, version mismatches, or non-deterministic outputs (missing random seeds).
- Validate against the report. Do the script's outputs match the numbers in the report? Spot-check headline tables, key effect sizes, sample sizes, p-values.
- Pre-specified vs exploratory. Is it clear which analyses were planned and which emerged from looking at the data? If not, flag it. If many tests were run, ask whether multiple-comparisons correction was applied.
- Assumption diagnostics. For each model:
- Linearity, normality (residuals), homoscedasticity, independence.
- Multicollinearity (VIF).
- Influential observations (Cook's distance, leverage).
- Was the appropriate model used (e.g., mixed effects for clustered data, robust SEs for heteroscedasticity)?
- Sensitivity analyses. Re-run key results with:
- Outliers excluded vs included.
- Alternative missing-data handling (listwise vs imputed).
- Alternative model specifications (with/without each control variable).
- Robust SE / non-parametric equivalent of any parametric test. How much do the conclusions change?
- Effect-size and uncertainty reporting. Are effect sizes reported with CIs, not just p-values? Are confidence intervals interpreted appropriately?
- Reproducibility. If you re-ran the analysis, would you get the same numbers? Are random seeds pinned?
Output
Return a validation memo:
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 · 117 lines · 93 tokens per session scan A 9f978d496cfa
stats-validator is an agent published in the GitHub repository Marazii/research-co-pilot (13 stars, last pushed 3mo ago), licensed MIT. It adds 93 tokens to every session and 1,347 once invoked, about $0.0005 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
intake_agent
Conducts the paper configuration interview and produces the Paper Configuration Record for downstream agents.
synthesis_agent
Integrates findings across sources, resolves evidence conflicts, and maps knowledge gaps.
argument_builder_agent
Constructs the papers core argument and logical reasoning structure.
data-engineering-prompt
Use this when: designing or fixing a data pipeline, choosing between batch/streaming/CDC, or picking table format and orchestration. Skip to: Protocol · Current stack · Remember.
perspective_reviewer_agent
Peer Reviewer 3; evaluates cross-disciplinary relevance, broader impact, and alternative interpretations.
critical_reviewer_agent
You are a Devil's Advocate reviewer whose job is to stress-test the paper's core arguments. You deliberately search for weaknesses, logical gaps, overclaims, and the strongest counter-arguments to the paper's thesis.