The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands to verify repository status and run project checks. It correctly identifies security risks by mandating the use of npm ci --ignore-scripts for the catalog repository, which prevents the execution of potentially malicious scripts during package installation. However, it also directs the agent to use the 'candidate project's own toolchain' for external tests, which relies on the agent's manual inspection to prevent the execution of malicious third-party code.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from external projects, including READMEs, documentation, and source code. This creates an attack surface where a malicious project could embed instructions to manipulate the agent's review or exfiltrate data.
Ingestion points: External project URLs, README files, and local source paths reviewed in SKILL.md (e.g., in the 'Review the actual artifact' section).
Boundary markers: The skill provides no explicit markers or delimiters to isolate third-party project content from the agent's core instructions.
Capability inventory: The agent can execute shell commands (via npm and project-specific toolchains), write to the local file system (to prepare patches), and interact with the GitHub API to create issues or pull requests.
Sanitization: The skill lacks automated sanitization of external content, relying entirely on the agent's instructions to 'inspect commands before running them' and 'review the diff for secrets.'
[EXTERNAL_DOWNLOADS]: The skill retrieves configuration, rules, and templates from official GitHub repositories owned by AppitStudio. These are recognized as legitimate vendor-owned resources necessary for the skill's primary function of preparing repository-compliant contributions.

awesome-jev-contributor