The Agent Skills Directory

[SAFE]: The skill performs routine development tasks such as collecting build artifacts, regenerate bindings, and rebuilding binaries using local scripts (node, bun). These operations are restricted to the local monorepo environment.
[SAFE]: Artifact downloads and uploads are conducted using the author's infrastructure at remotion.media. This aligns with the expected behavior for a vendor-maintained build tool.
[INDIRECT_PROMPT_INJECTION]: The skill instructions direct the agent to read data from other local skill files (../pr/SKILL.md and ../pr-name/SKILL.md) when preparing a pull request. This constitutes an attack surface for indirect prompt injection if those files were to contain malicious instructions. However, the risk is negligible as the files reside within the same controlled vendor repository.
Ingestion points: SKILL.md (reading ../pr/SKILL.md and ../pr-name/SKILL.md)
Boundary markers: None explicitly mentioned in the file-reading instructions.
Capability inventory: Shell command execution (node, bun) and repository management (git).
Sanitization: No explicit sanitization or filtering is mentioned for the content read from the external skill files.

update-ffmpeg