Home / agent-identity-governance
Agent Identity Governance
Nine skills that treat AI agents, service principals, workload identities, API keys and MCP connectors as identities with an inventory, owners, recertification, expiry and an off switch: one inventory across Entra ID, AWS IAM, GitHub and a hand-kept register; a review of Entra agent identities; quarterly recertification sheets per owner; an expiry radar for secrets, certificates, keys and tokens; a register check for MCP servers and connectors; a timeline of one agent's actions from audit exports; a kill switch runbook with a coverage check; a leaked credential response pack with hashed evidence; and a hash chain that makes an agent audit log tamper-evident and shows actions it never recorded. Standard-library scripts over saved exports, no network.
Install
In Claude Code, add the marketplace and install the plugin:
/plugin marketplace add basitalisandhu/claude-skills
/plugin install agent-identity-governance@claude-skills
Or copy the skill files into ~/.claude/skills/ from a clone:
git clone https://github.com/basitalisandhu/claude-skills
cd claude-skills
python3 install.py --user --only agent-identity-governance
Skills
| Skill | What it does | Scripts |
|---|---|---|
| agent-action-timeline | Reconstruct every action one agent or bot identity took, in time order, from saved Entra sign-in and audit logs, CloudTrail, the GitHub audit log and application logs in JSON lines, and flag bursts, unusual first-seen actions and actions outside a declared allow list. | yes |
| agent-audit-log-integrity | Seal an agent's audit log (JSON lines from a hook, MCP gateway or agent runtime) as a SHA-256 or HMAC hash chain, verify it later for edited, deleted, inserted or reordered lines and truncation against a stored head, and list actions the log never recorded. | yes |
| agent-kill-switch-runbook | Prepare the kill switch runbook to stop one agent identity fast: ordered steps to disable it, end sessions, remove credentials and grants, block it at the gateway and confirm it stopped in the audit log, each with a read-only verification command. | yes |
| agent-recertification | Run the quarterly recertification of agent and workload identities: one review sheet per owner to re-attest, saying what each identity can do, when it last acted and what changed, plus a tracking CSV. | yes |
| connector-register | Keep an approved register of the MCP servers, connectors and plugins your agents may use, and check every entry has an owner, purpose, data classification, permissions and review date. | yes |
| credential-expiry-radar | Find the secrets, certificates, access keys and tokens that have expired or soon will, across Entra app and Agent ID blueprint credentials, AWS IAM, GitHub fine-grained tokens and a CSV of other keys, bucketed at 7, 30 and 90 days with an action list per owner. | yes |
| entra-agent-id-review | Review the Entra identities and service principals your AI agents use, so each one has an owner or sponsor, only the permissions it needs (including those inherited from an Agent ID blueprint), current credentials, safe redirect URIs and Conditional Access cover. | yes |
| leaked-credential-response | Respond to a leaked agent key, API token or secret: what it could reach, what happened after the leak time, which credentials to replace first, and an evidence folder with SHA-256 hashes. | yes |
| nhi-inventory | Inventory every non-human or machine identity your agents and automations run as (service principals, IAM roles and users, GitHub apps and tokens), with owner, last use, credential count and oldest credential, and flag the ownerless and the unused. | yes |
Plugin README
Agent identity governance skills that compute from exports you already can take, each with a tested standard-library Python script. Install with /plugin marketplace add basitalisandhu/agent-identity-governance-skills and then /plugin install agent-identity-governance@agent-identity-governance-skills. Skills appear as /agent-identity-governance:<skill>, and Claude also invokes them on its own when a request matches a skill's description.
| Skill | Script | Use it to |
|---|---|---|
nhi-inventory | skills/nhi-inventory/scripts/nhi_inventory.py | merge Entra, AWS IAM, GitHub and register exports into one inventory of non-human identities with owner, last use and credential age, and flag ownerless, dormant and never-used identities |
entra-agent-id-review | skills/entra-agent-id-review/scripts/entra_agent_review.py | review Entra agent identities and agent-like apps for high-risk permissions, credential expiry, redirect URIs, owners and Conditional Access coverage |
agent-recertification | skills/agent-recertification/scripts/agent_recertification.py | prepare the quarterly recertification: one sheet per owner and a tracking CSV, flagging never attested, stale attestations, owners who left and removals not done |
credential-expiry-radar | skills/credential-expiry-radar/scripts/credential_expiry_radar.py | bucket secrets, certificates, access keys and tokens as expired, 7, 30 or 90 days, and list actions per owner |
connector-register | skills/connector-register/scripts/connector_register.py | check an MCP server, connector and plugin register for missing fields and overdue reviews, and against saved tool listings and permission manifests |
agent-action-timeline | skills/agent-action-timeline/scripts/agent_action_timeline.py | build one ordered timeline of an agent's actions from Entra, CloudTrail, the Azure activity log, GitHub and application logs, flagging bursts, first-seen actions and actions outside an allow list |
agent-kill-switch-runbook | skills/agent-kill-switch-runbook/scripts/kill_switch_runbook.py | write the ordered switch-off runbook for one identity with read-only verification commands, and check every credential and assignment has a step |
leaked-credential-response | skills/leaked-credential-response/scripts/leaked_credential_response.py | work out a leaked credential's reach, use after the leak and rotation order, and write a hashed evidence folder |
agent-audit-log-integrity | skills/agent-audit-log-integrity/scripts/audit_chain.py | seal an agent audit log as a SHA-256 or HMAC hash chain, verify it later for edited, removed, inserted or reordered lines and truncation, and list actions it never recorded |
Synthetic exports to try every skill on are in examples/ (each SKILL.md gives the command). They include an Entra Agent ID blueprint with two agent identities.
Requirements: Python 3.10 or newer on PATH as python3. The scripts read local files only: no network access, no third-party packages. Key ids are shown by their last four characters and secret values are never read.