basitalisandhu

Agent Identity Governance Skills — DevOps skill for Claude Code

DevOps community

Claude Code skills for agent identity governance: inventory the non-human identities your AI agents and automations run as (Entra service principals, managed identities, AWS keys and roles, GitHub App.

How to install Agent Identity Governance Skills

This entry records only its repository, not the path inside it, so there is no exact command to give. Open basitalisandhu/agent-identity-governance-skills and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agent Identity Governance Skills does

Claude Code skills for agent identity governance: inventory the non-human identities your AI agents and automations run as (Entra service principals, managed identities, AWS keys and roles, GitHub Apps, MCP connectors), review permissions, recertify quarterly, catch expiring credentials, trace what an agent did, switch one off fast. Offline.

Alternatives in DevOps

  • Kubeshark — Cluster-wide network observability for Kubernetes 11.8k ★
  • Open Agent Email — Open-source email infrastructure for AI agents: secure inboxes, REST and MCP APIs, SMTP, human approval workfl 2.1k ★
  • Azure Service Bus .net — Enterprise messaging with queues and topics 1.8k ★

README

Claude Code skills for agent identity governance: inventory, review, recertify, expire and switch off the non-human identities your AI agents and automations run as

**Eight skills that treat AI agents, service principals, workload identities, API keys and MCP connectors as identities: each needs an inventory entry, an owner, a periodic review, credentials that expire and an off switch. Offline, standard-library scripts over exports you save from Entra ID, AWS, GitHub and your own registers. Tested.**

What this is, who it is for, and why

agent-identity-governance-skills is a Claude Code plugin for the people who run identity and cloud security: Microsoft 365 and Entra ID administrators, AWS platform and security engineers, and whoever is asked "what can our agents actually do?" by an auditor or a board.

Every AI agent and automation acts as some identity: an Entra app registration or agent identity, a managed identity, an IAM user with an access key, an IAM role, a GitHub App or deploy key, or the API key behind an MCP connector. People get joiner, mover and leaver processes and quarterly access reviews; these identities usually get none of that. They keep their permissions after the pilot ends, their secrets after the owner leaves, and nobody can say quickly how to turn one off. Microsoft's own guidance on Entra Agent ID makes the same point: agent identities persist, accumulate access and need access reviews ([Microsoft 365 Copilot blog: security and governance innovations for Microsoft 365 Copilot and agents](https://techcommunity.microsoft.com/blog/microsoft365copilotblog/security-and-governance-innovations-for-microsoft-365-copilot-and-agents-from-ig/4476172)). When this pack was written, in October 2026, a GitHub search for non-human identity audit tooling returned no repositories; the vendor features that exist are consoles inside one product.

This pack does the governance work from files. Each skill reads exports you take with read-only commands (Gr