The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes several GitHub CLI (gh) commands to search for code, find repositories by topic, and interact with the GitHub API for repository enrichment.
[EXTERNAL_DOWNLOADS]: The skill clones external, unverified repositories from GitHub into a local temporary directory (.a5c/tmp/skill-discovery/) and fetches metadata from clawhub.ai to discover new workflows and plugins.
[REMOTE_CODE_EXECUTION]: The skill is designed to "assimilate" logic from external sources by parsing SKILL.md files and generating executable CommonJS (.cjs) files in the .a5c/processes/assimilated/ directory. Although these scripts are staged for review before being "promoted" to the library, the automated generation of executable code from untrusted remote content represents a significant security risk.
[PROMPT_INJECTION]: The skill processes markdown and text from arbitrary public repositories, creating a substantial attack surface for indirect prompt injection.
Ingestion points: Data is ingested from gh search results and the content of files (such as SKILL.md and plugin.json) within cloned repositories.
Boundary markers: No specific boundary markers or delimiters are used to isolate untrusted external content during processing.
Capability inventory: The skill has the ability to execute shell commands (gh), perform network operations (cloning, fetching), and write executable files to the local file system.
Sanitization: There is no evidence of sanitization or strict validation of the external content before it is parsed for procedural insights or used in code generation.

assimilate-popular-workflows