The Agent Skills Directory

[CREDENTIALS_UNSAFE]: The skill accesses a sensitive local file path ../secrets/github/pat.txt to retrieve a GitHub Personal Access Token (PAT).
[DATA_EXFILTRATION]: The retrieved PAT is used to authenticate a git push operation to a remote GitHub repository (github.com/farukzahra/agent-skills.git). While this targets the author's own infrastructure, the automated reading and transmission of authentication tokens is a high-risk pattern.
[COMMAND_EXECUTION]: The instructions direct the agent to execute a local PowerShell script (../agent-skills/scripts/sync-faruk-skills.ps1) and use the npx skills command to install or update packages.
[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input to create or update SKILL.md files. These files are later processed by the agent and can trigger script execution or network operations, creating a surface for indirect instruction injection.
Ingestion points: SKILL.md file content generated from user requests in the agent-skills repository.
Boundary markers: None identified.
Capability inventory: Local script execution (PowerShell), package management (npx), and git operations.
Sanitization: No specific sanitization or validation of the generated skill content is mentioned.

skills-sh-maintainer