The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires Node.js packages from the non-standard and unverified @earendil-works scope, such as @earendil-works/pi-coding-agent and @earendil-works/pi-ai, which are not recognized as trusted or well-known services.\n- [COMMAND_EXECUTION]: Instructions in references/reading.md direct the agent to execute shell commands such as pi --version and rg (ripgrep) to inspect the local environment and search through documentation.\n- [REMOTE_CODE_EXECUTION]: The skill uses node -e in references/reading.md to execute dynamic JavaScript logic at runtime for resolving package paths. This dynamic execution pattern poses a risk if input variables are manipulated.\n- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection. Ingestion points: The agent is instructed to read docs/sdk.md and dist/index.d.ts from the local project (identified in references/reading.md). Boundary markers: Absent; no delimiters or ignore instructions are provided for the content of these files. Capability inventory: The skill possesses the ability to execute shell commands (pi, node, rg) and generate code for SDK integrations. Sanitization: Absent; content from the documentation and type definitions is processed directly into the agent context without validation or filtering.

pi-sdk