The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides instructions to execute various shell commands for library management and debugging.
Evidence: bun run opensrc:sync and bun run opensrc:use <package> in SKILL.md are used to synchronize and fetch external libraries.
Evidence: The opencode CLI tool is executed with various arguments in references/opencode-skill-testing.md for skill loading verification.
[DYNAMIC_EXECUTION]: The skill includes a specific pattern for executing processes through a language interpreter wrapper.
Evidence: references/opencode-skill-testing.md contains the command perl -e 'alarm 120; exec @ARGV' opencode ..., which uses Perl's exec function to spawn the opencode process with a timeout. This is a form of dynamic process management.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze external data (source code from third-party libraries) which represents an attack surface for indirect prompt injection.
Ingestion points: Files located within opensrc/repos/github.com/ across multiple libraries (Effect, TanStack, TRPC, etc.) as detailed in SKILL.md.
Boundary markers: None explicitly defined in the instructions to prevent the agent from following instructions found within the code comments or documentation of the analyzed libraries.
Capability inventory: The skill has the capability to execute shell commands (bun, opencode) and read arbitrary files in the mirrored repository paths.
Sanitization: There is no mention of sanitization or filtering for the external content being read.

debugging-with-opensrc