The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: Documents the installation of supporting skills using npx from the author's namespace (HKUDS/CLI-Anything).- [REMOTE_CODE_EXECUTION]: Instructs the agent to perform environment setup that includes downloading and executing code via pip for Python dependencies and npx for Node.js-based skill management.- [COMMAND_EXECUTION]: Spawns shell processes for repository cloning (git clone), test execution (pytest), and the invocation of custom internal scripts such as repl_skin.py and preview_bundle.py.- [DYNAMIC_EXECUTION]: Employs a dedicated script, skill_generator.py, to dynamically generate new skill files and harnesses based on the analysis of target applications.- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process data from untrusted external repositories and local directory paths.
Ingestion points: User-supplied repository URLs and file paths provided to build and refine commands.
Boundary markers: Absent; there are no specific instructions or delimiters to isolate repository content from the agent's instructions.
Capability inventory: Includes shell command execution (cloning, installing, testing) and file write permissions to implement harnesses.
Sanitization: No filtering or sanitization of ingested content is described.

cli-anything-generator