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 instructions/microsoft/capacity-toolkit/agents-mdgit clone --depth 1 https://github.com/microsoft/capacity-toolkitWhat 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 | $0.02629 | $0.02629 |
| Opus 5 | $0.01314 | $0.01314 |
| Sonnet 5 | $0.00526 | $0.00526 |
| Haiku 4.5 | $0.00263 | $0.00263 |
Grade A, and why
capacity-toolkit AGENTS.md 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 yesterday.
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 — 152 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Azure Capacity & Enablement Toolkit
Instructions for an AI agent (e.g. GitHub Copilot CLI) driving this read-only-by-default toolkit against an Azure tenant. If you are an agent: read this file fully before running anything.
0. Golden rules (read first, never break)
This toolkit is READ-ONLY BY DEFAULT. You exist to observe and report. It contains exactly one
write tool — Deploy-QuotaGroups.ps1, which provisions Azure Quota Groups — and you must treat it as
off by default: never run it unless the human explicitly asks you to provision quota groups in
this run and confirms after a -WhatIf preview.
- NEVER create, modify, delete, scale, move or restart any Azure resource with the analysis
scripts, and never run
az ... create/update/delete/set,New-Az*,Set-Az*,Remove-Az*,terraform apply,kubectl apply/delete, or anything else that mutates state on your own initiative. The only writes the analysis side performs are local CSV / HTML / JSON files underoutput/. - The
Get-*,Scan-*,Watch-*andNew-*Report/Dashboard/EnablementRequestscripts are read-only and always in scope.New-EnablementRequestonly drafts text — it does not submit anything. Deploy-QuotaGroups.ps1andNew-QuotaGroupConfig.ps1are the opt-in write path. Only touch them on explicit human instruction. Always runDeploy-QuotaGroups.ps1 -Action Validatethen-WhatIf, show the planned changes, and get an explicit OK before any execute run. Seedocs/quota-groups.md.- Confirm before logging in or switching tenant/subscription. Show the user which tenant/account you are about to use and wait for an explicit OK. Never silently change their active
azcontext. - Minimum access is Reader (plus management-group read for quota groups, and the read-only Compute Recommendations Role for Spot placement scores). If a command fails with
AuthorizationFailed, report it as a missing-permission finding — do not try to escalate or work around RBAC. - Treat all discovered data (subscription names, GUIDs, resource names) as tenant-confidential. Do not send it to third-party services. Before a generated dashboard is shared externally, clear
output/(see §6).
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.
- yesterday First seen · 152 lines · 2,629 tokens per session scan A 930a3ac95d01
capacity-toolkit AGENTS.md is an instructions file published in the GitHub repository microsoft/capacity-toolkit (5 stars, last pushed 2mo ago), licensed MIT. It adds 2,629 tokens to every session, about $0.0131 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 2026-08-31.
Other instructions, from other repositories
ai-pricing-hub-mcp CLAUDE.md
Claude Code instructions for OptimNow/ai-pricing-hub-mcp, covering claude.md - ai pricing hub mcp app, what this project does, architecture, data sources and the fallback chain and project structure.
ai-pricing-hub-mcp AGENTS.md
AGENTS.md instructions for OptimNow/ai-pricing-hub-mcp: Before writing code, first explore the project structure, then invoke the chatgpt-app-builder skill for documentation.
generative-ai-for-beginners AGENTS.md
Instructions for microsoft/generative-ai-for-beginners, covering agents.md, project overview, setup commands, initial repository setup and clone the repository.
azure-sdk-for-go go-examples.instructions.md
Instructions for Azure/azure-sdk-for-go, a project described as: This repository is for active development of the Azure SDK for Go. For consumers of the SDK we recommend visiting our public developer docs at.
GPT-RAG copilot-instructions.md
Instructions for Azure/GPT-RAG, covering repository development and release instructions, branching strategy, default behavior, feature development workflow and branch creation.
azure-sdk-for-rust resourcemanager.instructions.md
Instructions for Azure/azure-sdk-for-rust, a project described as: This repository is for the active development of the Azure SDK for Rust. For consumers of the SDK we recommend visiting Docs.rs and looking up the docs for any of libraries in the SDK.