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 skills add ur-grue/autopunk-media-skills --skill data-outlier-findergit clone --depth 1 https://github.com/ur-grue/autopunk-media-skillsWrote 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/skills/ur-grue/autopunk-media-skills/data-outlier-finder)<a href="https://agentmods.dev/skills/ur-grue/autopunk-media-skills/data-outlier-finder"><img src="https://agentmods.dev/badge/skills/ur-grue/autopunk-media-skills/data-outlier-finder/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/skills/ur-grue/autopunk-media-skills/data-outlier-finder"><img src="https://agentmods.dev/badge/skills/ur-grue/autopunk-media-skills/data-outlier-finder.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector pass
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 | $0.00032 | $0.01444 |
| Opus 5 | $0.00016 | $0.00722 |
| Sonnet 5 | $0.00006 | $0.00289 |
| Haiku 4.5 | $0.00003 | $0.00144 |
Grade A, and why
data-outlier-finder 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 13d 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 — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Data outlier finder
What this skill does
Identifies unusual values, unexpected patterns, and potential stories hidden in a dataset by systematically checking for statistical outliers and contextual anomalies.
When to use this skill
- You have a dataset and want to know where the stories are before you start reporting
- An editor asks "what's surprising in this data?" and you need a structured answer
- You are fact-checking a claim and want to verify whether the cited figure is genuinely unusual
- You received a data dump (FOI response, leaked spreadsheet, public database) and need to triage it for newsworthy patterns
What you need to provide
Required:
- The dataset or a representative sample (paste the data, describe the columns, or share summary statistics)
- What the data measures and where it comes from
Optional:
- What you expect to find (so the assistant can flag deviations from your assumptions)
- Known context that might explain outliers (e.g., "2020 data will be distorted by COVID")
- Whether you want statistical outliers only or also contextual anomalies (values that are technically valid but editorially surprising)
How the Assistant Approaches This
- Identifies the key numeric columns and their expected ranges
- Flags statistical outliers — values more than 2 standard deviations from the mean, or in the top/bottom 5% of the distribution
- Checks for contextual anomalies — sudden changes between time periods, values that contradict known patterns, or categories that break expected rankings
- For each outlier, states: what the value is, why it's unusual, and what reporting question it raises
- Prioritizes findings by editorial potential — leads with the outliers most likely to be stories, not just noise
Output format
A numbered list of findings, ordered by editorial potential. Each finding includes:
- The value and where it appears in the dataset
- Why it's unusual (statistical context or comparison)
- The reporting question it raises — what a journalist should check next
What ships with it
1 file beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
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.
- 13d ago First seen · 90 lines · 32 tokens per session scan A 32378eace458
data-outlier-finder is a skill published in the GitHub repository ur-grue/autopunk-media-skills (32 stars, last pushed 12d ago), licensed MIT. It adds 32 tokens to every session and 1,444 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 skills, from other repositories
entity-graph
Resolve entities across an investigation's documents into a Neo4j graph after a mandatory human review gate. Layer-2, operator-tier, Docker-gated -- run it to deduplicate people/orgs across the corpus, decide the "maybe" matches yourself, then write the resolved network to Neo4j for queryable analysis.
analysis-recipe
This skill should be used when the user asks to "run the analysis recipe", "run the 13-point checklist", "run the per-source investigative pass", "analyze this FOIA/audit-log source for out-of-state / immigration / pretext / mega-users / co-travel", "roll up findings across multiple sources/agencies", or "test the…
pii-sweep
This skill should be used when the user asks to "sweep PII exposure in a FOIA free-text field", "quantify how much PII a free-text reason/narrative column exposed", "run the authoritative PII tally", "count names / SSNs / A-numbers / DOBs in a reason column weighted by row count", "separate officials named for…
entity-crossref
Cross-reference an investigation's resolved entities against the investigator's own corpus (FOSS default, zero external data) and, opt-in, sanctions/PEP watchlists, using a local yente + OpenSearch stack. Layer-2, operator-tier, Docker-gated -- run it after building the entity graph to screen the resolved people/orgs…
ingest
This skill should be used when the user asks to "ingest a PDF / document", "extract text from a scanned PDF preserving page and bounding-box provenance", "convert a FOIA PDF to DoclingDocument keeping citations", "OCR a scanned document but decide native-text-vs-re-OCR first", "turn a PDF into provenance-preserving…
redact-output
This skill should be used when the user asks to "redact PII before publishing", "redact third-party names to initials in a findings note", "redact a note before it goes in the vault", "mask suspect / person-of-interest / minor names that landed in the reason fields", "keep the full un-redacted exhibit local / out of…