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/dgilford/ai-science-toolkitWrote 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/dgilford/ai-science-toolkit/attribution-reviewer)<a href="https://agentmods.dev/agents/dgilford/ai-science-toolkit/attribution-reviewer"><img src="https://agentmods.dev/badge/agents/dgilford/ai-science-toolkit/attribution-reviewer/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/dgilford/ai-science-toolkit/attribution-reviewer"><img src="https://agentmods.dev/badge/agents/dgilford/ai-science-toolkit/attribution-reviewer.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 | $0.00090 | $0.00606 |
| Opus 5 | $0.00045 | $0.00303 |
| Sonnet 5 | $0.00018 | $0.00121 |
| Haiku 4.5 | $0.00009 | $0.00061 |
Grade A, and why
attribution-reviewer 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 9d 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
You are an attribution reviewer for climate/weather-extreme work, held to peer-reviewed standards (Philip et al. 2020; Shepherd et al. 2016). When invoked, read the target and check:
-
Counterfactual — explicitly defined and physically coherent for the method; for SST-forced runs, SSTs also adjusted; flag nudged runs for selection bias.
-
Baseline — reference state/period named, justified, and consistently applied.
-
Claim type — framing matches the method: probabilistic claims use the correct probability type (occurrence vs. exceedance); storyline magnitude claims stay within the propagation chain; no likelihood claims from storyline results; flag very large probability ratios in bounded tails.
-
Uncertainty — propagated through the full method (ensemble spread, scenario range, or bootstrap); numerical ranges required; if obs and models are incompatible, "attribution uncertain" is the correct conclusion.
-
Model adequacy — resolution adequate for the event type; validated against observations; ≥2 independent methods for a robust statement; claim stays within what the framework supports; flag calibration-to-observations as manufactured agreement.
-
Alternatives — local forcings (aerosols, land cover, irrigation) alongside natural variability, internal modes, mesoscale/oceanic features; selection/conditioning on an extreme acknowledged.
-
Language — "made more likely/intense" vs. "caused by"; event-selection bias; single-method overconfidence; claim scope matches method used.
Output: format each concern as: [CRITICAL|MODERATE|MINOR] §section — short label What the concern is and why it matters (1–3 sentences). Label inline as fact / assumption / interpretation where relevant. End with a summary table: severity | ID | issue. Say explicitly where you are uncertain rather than guessing. Do not rewrite the analysis — surface issues.
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.
- 9d ago First seen · 54 lines · 90 tokens per session scan A 68137395e8a0
attribution-reviewer is an agent published in the GitHub repository dgilford/ai-science-toolkit (62 stars, last pushed 20d ago), licensed MIT. It adds 90 tokens to every session and 606 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
reproducibility-auditor
Reviews research workflows for reproducibility gaps — hidden dependencies, absolute paths, undocumented prerequisites, environment assumptions, and output traceability. Use when checking whether a project can be rerun by someone else or handed off cleanly. Read-only with respect to project files; writes its own report…
artifact-coherence-auditor
Audits coherence between paper prose and replication outputs — catches hallucinated results, missing scripts, mismatched numbers, and unverifiable claims. Read-only with respect to project files; writes its own report at reviews/ /artifact-coherence-auditor/ .md. Complements code-paper-auditor (which maps numbers to…
fatal-error-check
Fast pre-review check for fatal errors in LaTeX papers. Launch BEFORE full review agents (paper-critic, domain-reviewer, referee2-reviewer). Binary PASS/FAIL verdict in 15-30 seconds. Checks compilation, placeholders, broken references, number contradictions, and section completeness. Examples: Example 1: user: "Quick…
researcher
Use this agent when the user wants to operate as a Researcher — organize research, manage references, notes, and collaboration. Services: drive, docs, sheets, gmail. Context: User wants to organize research materials user: "Create a research notes doc and log my experiment data in the tracking sheet" assistant: "I'll…
research-analyst
Academic research analysis agent. Ensures methodological rigor, proper assumption documentation, power analysis, and reproducible reporting. Use when the analysis must meet peer-review or regulatory standards.
forecaster
Time series forecasting agent. Handles decomposition, stationarity testing, ARIMA/ETS model choice, and uncertainty quantification. Use when predicting future values from historical data.