The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill provides instructions to download an installation script from the vendor's official domain at https://composio.dev/install. This is a documented procedure for setting up the Composio CLI environment.
[REMOTE_CODE_EXECUTION]: The installation command curl -fsSL https://composio.dev/install | bash pipes a remote script directly into the shell. While this is a common installation pattern for CLI tools, it involves executing remote code. The source is the official domain of the service provider.
[COMMAND_EXECUTION]: The skill utilizes several CLI commands to perform its functions:
composio execute: Runs specific tool actions defined by the service.
composio run --file ./workflow.ts: Executes local TypeScript files to automate multi-step workflows.
composio proxy: Allows authenticated API calls to external URLs via the platform's authentication management.
[COMMAND_EXECUTION]: The skill processes data fetched from external SaaS applications (such as Slack, GitHub, or Gmail). This creates an ingestion surface for indirect prompt injection where instructions embedded in external content could potentially influence the agent's logic.
Ingestion points: Data retrieved via composio execute, composio proxy, and composio listen (triggers).
Boundary markers: None explicitly defined in the instructions to separate external data from agent instructions.
Capability inventory: File system access (via --file), network proxying, and arbitrary tool execution across integrated platforms.
Sanitization: No specific sanitization or validation logic for the content of retrieved data is described.

composio-cli