The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill is designed to fetch documentation from external sources using git clone for repositories and curl for web-based documentation. These operations are initiated by the operator to populate a local documentation mirror.
[COMMAND_EXECUTION]: The skill utilizes shell scripts (create-docs-skill.sh and a generated generate-docs-skill.sh) to automate the scaffolding and updating of documentation skills. These scripts perform standard file system operations, such as creating directories, copying files, and generating hash manifests for verification.
[INDIRECT_PROMPT_INJECTION]: The skill implements a proactive 'neutralization' strategy to mitigate risks associated with ingesting untrusted external data. Mirrored files that match common instruction or configuration patterns (e.g., SKILL.md, Dockerfile, package.json, and various agent instruction files like AGENTS.md) are automatically renamed with a .source suffix. This prevents the agent's platform from accidentally discovering or interpreting these files as active instructions or configuration. Additionally, the generated skill includes explicit warnings to treat mirrored content as untrusted.
[DYNAMIC_EXECUTION]: The documentation mirroring script uses an embedded Python one-liner to perform text normalization for consistency checks. The Python code is static, utilizes the standard library, and operates on local files within the mirroring process.

docs-skill-builder