context-setup
Pass
Audited by Gen Agent Trust Hub on Oct 10, 2026
Risk Level: SAFECOMMAND_EXECUTIONPERSISTENCEINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell and Node.js scripts (install-context.sh, install-context.js) located in the plugin's root directory to perform setup and verification tasks.
- [COMMAND_EXECUTION]: User-selected profile input is interpolated directly into shell command arguments (--profile={profile}), which represents a potential surface for command injection.
- [PERSISTENCE]: The skill modifies the user's home directory (~/.claude/) to store global configuration and standards, allowing these settings to persist across different projects and sessions.
- [INDIRECT_PROMPT_INJECTION]: The skill reads and displays metadata from manifest files, which could be manipulated to inject instructions into the agent's output.
- [INDIRECT_PROMPT_INJECTION]: Ingestion points: Project-local and global (.claude/.context-manifest.json) manifest files.
- [INDIRECT_PROMPT_INJECTION]: Boundary markers: Absent for the status display.
- [INDIRECT_PROMPT_INJECTION]: Capability inventory: The skill is capable of file writing (printf), shell execution (bash), and Node.js execution (node) as defined in SKILL.md.
- [INDIRECT_PROMPT_INJECTION]: Sanitization: No explicit validation or escaping is applied to the manifest data before it is presented to the user.
Audit Metadata