The Agent Skills Directory

[DYNAMIC_EXECUTION]: The skill is designed to generate custom bash scripts at runtime by combining a static library with logic authored by the AI.
Evidence: SKILL.md defines a process where the agent authors script "stages" and combines them with a library provided in template.sh to create a new executable script.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from the local repository to determine setup requirements, creating a vector where malicious repository content could influence the AI's generated script logic.
Ingestion points: SKILL.md instructs the agent to read .env, README, docker-compose*, and .github/workflows/* files from the repository.
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are specified when processing these files.
Capability inventory: Generated scripts have the capability to write to .env files, open URLs in the user's browser, and interact with the GitHub CLI (gh) to set secrets and variables.
Sanitization: No sanitization or validation of ingested file content is performed before logic generation.
[COMMAND_EXECUTION]: The skill instructs the user to modify file permissions and execute the generated shell scripts.
Evidence: SKILL.md Step 3 explicitly provides the command chmod +x to make the generated scripts executable by the user.

wizard