cyrus-setup-github
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONDATA_EXFILTRATIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill executes multiple shell commands to check authentication status, configure git identity, and interact with the GitHub API using
ghandcurl. - [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface for indirect prompt injection through the interpolation of user-supplied data into shell commands.
- Ingestion points: User-provided values for git
name,email, and the GitHub AppHOMEPAGE_URLin SKILL.md. - Boundary markers: Absent.
- Capability inventory: Shell command execution (
git config,printf), file system modification, and network operations. - Sanitization: Absent; user inputs are placed directly into command strings without escaping or validation.
- [DYNAMIC_EXECUTION]: Generates and runs a Node.js script at runtime to handle cryptographic operations.
- The skill uses
node -eto execute a script that reads a private key and generates a JSON Web Token (JWT) for GitHub App authentication. - [DATA_EXFILTRATION]: Accesses and manages sensitive credential files on the local filesystem.
- Interacts with
~/.cyrus/.envand~/.cyrus/github-app.pemto store and retrieve webhook secrets and RSA private keys. It includes a specific warning to the agent to avoid reading these files into the conversation context to prevent accidental exposure. - [EXTERNAL_DOWNLOADS]: Communicates with GitHub's official API endpoints to exchange manifest codes for app credentials and to manage installations.
Audit Metadata