The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes content from third-party instruction files and search results, creating a surface for indirect prompt injection attacks.
Ingestion points: Reads SKILL.md files of installed skills to extract metadata and parses output from npx skills find.
Boundary markers: Does not use specific delimiters to isolate external text, relying instead on the agent's adherence to the audit policy in references/security-filter.md.
Capability inventory: Accesses Bash for installation and Write/Edit for updating project routing files like AGENTS.md.
Sanitization: Uses a logical security filter to identify and block malicious patterns in external skills before installation.
[DYNAMIC_EXECUTION]: Shell commands are built at runtime using identifiers obtained from an external registry search.
Evidence: The skill generates commands such as npx skills add <owner/repo> --skill <skill-name> where placeholders are filled from dynamic search results.
[EXTERNAL_DOWNLOADS]: The skill's primary function is to download and install external code components (agent skills) from a remote registry.
Evidence: Explicitly uses npx skills add to fetch resources from the skills.sh ecosystem.
[COMMAND_EXECUTION]: Uses local PowerShell and Bash scripts to scan the project environment.
Evidence: Executes ./assets/detect-stack.ps1 and ./assets/detect-stack.sh to determine the project's technology stack.

project-onboarding