The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/configure-keyapi-auth.mjs modifies user shell configuration files (such as .bashrc, .zshrc, and PowerShell profiles) to persist the KEYAPI_TOKEN environment variable. This modification of system startup scripts is a persistence mechanism that can affect the security and behavior of the user's terminal environment.
[DATA_EXFILTRATION]: The script scripts/configure-keyapi-auth.mjs reads the contents of sensitive shell configuration files (.bashrc, .zshrc, etc.) to check for existing credentials. Accessing these files can expose other environment variables or system secrets stored in the user's profile.
[DATA_EXFILTRATION]: The script scripts/keyapi-api.mjs provides the capability to read any local file through the --body-file and --image-file arguments and send its content to the api.keyapi.ai domain. This functionality can be misused to exfiltrate sensitive files, such as SSH keys, private keys, or environment files, if the agent is directed to read from those paths.
[EXTERNAL_DOWNLOADS]: The script scripts/search-keyapi-docs.mjs fetches documentation from docs.keyapi.ai to dynamically resolve API endpoints at runtime.
[PROMPT_INJECTION]: The skill ingests untrusted data from the Threads API through scripts/keyapi-api.mjs and lacks explicit boundary markers or sanitization before incorporating this data into its context. This creates a surface for indirect prompt injection. Ingestion point: API responses in keyapi-api.mjs. Boundary markers: Absent. Capability inventory: Script execution, file system access, and network operations. Sanitization: Absent.

keyapi-threads