heating the forge...

@forgecat/microsoft-azure_skills-entra-app-registration — Security Risk Report | ForgeCat

microsoft-azure_skills-entra-app-registration

Security risk report for @forgecat/microsoft-azure_skills-entra-app-registration v0.0.4

Low

Static scanner flagged safe level: profile metadata attributes Microsoft as author with reference to official Microsoft azure-skills repository (https://github.com/microsoft/azure-skills).
Content is educational documentation for standard Microsoft Entra ID app registration workflows; no supply-chain poisoning or typosquatted dependencies detected.
All external references point to official Microsoft documentation (learn.microsoft.com, graph.microsoft.com, jwt.ms) and legitimate open-source libraries (MSAL, Azure SDK).

Low

Content is purely educational and descriptive: it documents OAuth 2.0 flows, app registration steps, and authentication patterns without instructing the AI to perform malicious actions.
No prompt injection, role hijacking, or system prompt leakage attempts detected; no instructions to read credentials from ~/.ssh, ~/.aws, .env, or exfiltrate data.
Security best practices section (auth-best-practices.md) actively recommends defensive patterns (managed identity, least privilege, token validation, HTTPS-only) rather than weakening security defaults.

Details

Findings

Invisible characters detected
skills/entra-app-registration/references/cli-commands.md · SKILL-011
Invisible characters detected
skills/entra-app-registration/references/first-app-registration.md · SKILL-011

Evidence

"Use managed identities and Azure RBAC in production. Reserve DefaultAzureCredential for local development only."
"Never hardcode secrets... Use environment variables, Azure Key Vault, or managed identity."
"Always validate issuer, audience, expiration" and "Use HTTPS only" in security best practices table.

Low

Profile declares no tools, hooks, or MCP servers; it is pure markdown documentation with code examples and CLI command references.
No file_write, file_delete, shell, or other high-risk tool categories are requested; the skill is informational only.
CLI commands shown (az ad app create, az ad app permission add, etc.) are standard Azure CLI operations documented by Microsoft and require explicit user invocation.

Details

Evidence

Profile metadata: 'Dependencies: None' and no tool definitions in SKILL.md.
CLI examples are presented as reference documentation, not as auto-executing hooks.
All commands require user to set variables (APP_ID, CLIENT_ID, TENANT_ID) and run them manually.

Low

No MCP servers are defined in the profile; the skill is pure markdown documentation with code examples.
No binary execution, network access, or filesystem operations are configured.
References to external tools (Azure CLI, MSAL libraries) are standard, well-known, and documented by Microsoft.

Details

Evidence

Profile metadata lists 'MCP Tools and CLI' section but defines no MCP server configurations.
All tool references are to official Microsoft packages (Microsoft.Identity.Client, @azure/msal-browser, msal, Azure CLI).