microsoft/cat-agent-skills is a static website that catalogs reusable instruction sets and related packages for AI agents. People use it to search, filter, rate, and download skills for Cowork, Copilot Studio, and Scout, along with Copilot plugins and Scout automations. The catalogue entries are the skills, instructions, plugins, and settings displayed by the site.
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 microsoft/cat-agent-skills --skill denial-classifygit clone --depth 1 https://github.com/microsoft/cat-agent-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/microsoft/cat-agent-skills/denial-classify)<a href="https://agentmods.dev/skills/microsoft/cat-agent-skills/denial-classify"><img src="https://agentmods.dev/badge/skills/microsoft/cat-agent-skills/denial-classify/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/microsoft/cat-agent-skills/denial-classify"><img src="https://agentmods.dev/badge/skills/microsoft/cat-agent-skills/denial-classify.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 | $0.00213 | $0.01599 |
| Opus 5.5 | $0.00085 | $0.00640 |
| Sonnet 5.5 | $0.00043 | $0.00320 |
| Haiku 4.5 | $0.00021 | $0.00160 |
Grade A, and why
denial-classify 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 2d 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 — 46 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Denial Classify
Purpose
Establish what kind of denial this is, from the denial document itself, before anything else runs.
When to use
Use at the start of every denial/appeal run, on receipt of a denial letter or remittance advice (X12 835 or its paper equivalent) - typical asks are "we got a denial from Contoso dated 8/20", "what kind of denial is this", "read this EOB and start the appeal".
When NOT to Use
Do NOT use when: the denial has already been classified on the contract (go straight to policy-retrieve); the question is about a pre-service prior-authorization denial - use the prior-auth-assist plugin instead; the ask is to draft or reword the appeal letter - use appeal-draft instead; the ask is to find chart documentation - use evidence-assemble instead; the ask is to interpret payer policy language or compute the deadline - use policy-retrieve or deadline-track instead; or no denial document is available - there is nothing to classify from and guessing is not permitted.
Inputs
Denial letter or remittance advice (X12 835 or paper equivalent), read with the session's file-reading tool (e.g. ReadFileContent for a OneDrive/SharePoint file); original claim and coding (837) when supplied; config/payer-appeal-windows.json for the appeal level.
Steps
- Read the denial document and copy verbatim: payer (835 N1PR), claim_id (CLP01), claim_value (CLP03), paid amount (CLP04), adjustment group, reason code and amount (CAS01/02/03), remark codes (claim-level MIA/MOA or service-level LQHE - record which segment), and original_code (837 SV1/SV2; if no 837 is supplied, the 835 SVC echo, disclosed as such).
- Set denial_date to the date printed on the letter/remittance (835: DTM*405). If the document carries more than one candidate date and they differ, leave denial_date unset and escalate - the rule for which date starts the appeal window is [TO CONFIGURE: payer-specific rule for which date starts the appeal window]. Set as_of to today's date (the date this classification was performed).
- Set recoverable_value from the remittance amounts only (CLP03 minus CLP04, cross-checked against CAS03); if they disagree, leave it unset for a human.
- Classify the denial type from the extracted codes (e.g. medical necessity, timely filing, coding, authorisation missing) - never from what the requester says the denial is.
- Set the appeal route to the payer's configured level from config/payer-appeal-windows.json payers[payer].level; if the payer is not configured, record "not configured" and let deadline-track escalate.
- Record the source document (name and date) and the source segment for every field, write denial{} to the contract as the entry hop, and hand off to policy-retrieve.
What ships with it
3 files 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.
- 2d ago First seen · 46 lines · 213 tokens per session scan A 49ec586c7f1b
denial-classify is a skill published in the GitHub repository microsoft/cat-agent-skills (101 stars, last pushed yesterday), licensed MIT. It adds 213 tokens to every session and 1,599 once invoked, about $0.0009 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 2026-11-06.
denial-classify compared with skills from other repositories
| Mod | Type | Stars | Last push | Licence | Scan | Tokens | Works with |
|---|---|---|---|---|---|---|---|
| denial-classifymicrosoft/cat-agent-skills | Skill | 101 | yesterday | MIT | A | 213 | Claude Code, Codex |
| sector-rotationHKUDS/Vibe-Trading | Skill | 38k | today | MIT | A | 39 | Claude Code, Codex |
| strategy-pivot-designertradermonty/claude-trading-skills | Skill | 3.7k | today | MIT | A | 28 | Claude Code, Codex |
| twitter-readerhimself65/finance-skills | Skill | 3.5k | 4d ago | MIT | A | 110 | Claude Code |
| energy-procurementhashgraph-online/awesome-codex-plugins | Skill | 1.5k | today | Apache-2.0 | A | 93 | Claude Code, Codex |
| airdrop-hunting-guidenirholas/three.ws | Skill | 271 | today | Apache-2.0 | A | 54 | Claude Code, Codex |
| multi-expert-analyzerdigoal/blog | Skill | 8.6k | 4d ago | GPL-2.0 | A | 292 | Codex |