The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes multiple bash and python scripts from its local directory to perform discovery, validation, and linting of the project (e.g., run-discovery.sh, run-validation.sh). It also manages project dependencies via pip, npm, and bundle, and can perform system-level tasks such as updating bash via Homebrew on macOS.
[EXTERNAL_DOWNLOADS]: The skill downloads official Telnyx SDK packages from public registries (pypi.org, npmjs.com) and communicates with Telnyx API endpoints (api.telnyx.com) for account verification and integration testing. These downloads are conducted from trusted, well-known services.
[PROMPT_INJECTION]: The skill presents an attack surface for indirect prompt injection (Category 8) because it autonomously modifies project source code based on data ingested during its discovery phase. Evidence chain: 1. Ingestion points: The skill scans all project files in the specified root directory. 2. Boundary markers: Absent; there are no explicit delimiters or instructions to the agent to disregard embedded instructions within user project files. 3. Capability inventory: The skill has authority to write to the file system, execute shell commands, and perform network operations. 4. Sanitization: Absent; the logic for code transformation is driven by the results of the project scan without a verification step for external content.

telnyx-twilio-migration