The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The skill's primary function is to locate and install new executable code into the agent's environment. It uses the bin/add-skill utility to download skill packages from external repositories.
[COMMAND_EXECUTION]: The skill executes several system utilities to analyze repository state and manage the search process, including ls, jq, grep, and npx for CLI-based searches.
[EXTERNAL_DOWNLOADS]: The skill fetches data from multiple external surfaces to find candidates, including the GitHub API, official organization indexes, and the skills.sh search directory. It also downloads repository tarballs during the installation phase.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted text descriptions from the open web and third-party catalogs. While it includes defensive instructions to prevent the agent from obeying embedded commands, this ingestion constitutes a potential attack surface.
Ingestion points: External skill names and descriptions fetched from npx skills find, bin/add-skill --list, and skills.sh search results.
Boundary markers: The skill contains an explicit instruction to treat fetched descriptions as untrusted data and to ignore any instructions embedded within them.
Capability inventory: The skill can write to the file system, modify configuration files (aeon.yml), update dependency locks (skills.lock), and initiate network downloads.
Sanitization: The skill relies on the agent's instruction-following capabilities and a secondary security scan (scripts/skill-scan.sh) triggered during the installation process.

search-skill