| name | zworkforce-policy-audit |
| description | Audit zWorkforce tenant policy, RBAC, scopes, API keys, proxy identity, approval gates, agent tool grants, signed skills, MCP allowlists, SSRF defenses, shell execution limits, memory isolation, and audit-chain integrity. |
zWorkforce Policy Audit
Assume policy errors are production blockers until proven otherwise.
Audit Path
- Identify tenant, actor, roles, scopes, agent, skill IDs, mutating intent, and
requested tools.
- Trace enforcement across API, database, engine, tools, policy, identity,
secret store, and audit code.
- Verify deny precedence and fail-closed behavior.
- Confirm approval behavior distinguishes
requires_approval_for_mutations,
required_approvals, and approval_tools.
- Check signed remote skill behavior and registry allowlists.
- Validate that audit events capture enough evidence without leaking secrets.
Evidence Files
Start with:
docs/THREAT-MODEL.md
docs/IDENTITY.md
docs/SECRET-MANAGEMENT.md
docs/MCP.md
zworkforce/policy.py
zworkforce/tools.py
zworkforce/identity.py
zworkforce/skills.py
- related tests under
tests/
Output
Classify issues as blocking, high, medium, or low. Include a concrete exploit or
failure mode for blocking and high findings.