The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Installs the Node.js package @bitbasti/chat-adapter-webex@0.1.0. This is a third-party dependency not hosted within the primary vendor's namespace.
[COMMAND_EXECUTION]: Executes shell commands via pnpm to build the project and run integration tests (pnpm run build, pnpm exec vitest).
[PROMPT_INJECTION]: The skill exhibits an attack surface for indirect prompt injection by interpolating untrusted data from user prompts into the system configuration.
Ingestion points: The nc:prompt directive in SKILL.md captures sensitive strings (bot_token, webhook_secret) directly from user input.
Boundary markers: No specific delimiters or instructions are provided to the agent to treat the collected input as data only, increasing the risk of the agent interpreting input content as instructions.
Capability inventory: The skill is capable of performing file system writes to .env and executing arbitrary shell commands through the build and test directives.
Sanitization: There is no evidence of input validation or sanitization for the captured secret tokens before they are persisted to the environment.

add-webex