The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill scans untrusted project manifest files (such as package.json, Cargo.toml, and Makefile) to automatically identify build, test, and lint commands. An attacker could place malicious instructions in these files to attempt to influence the agent's behavior or trick the user into executing harmful commands during the configuration confirmation step.
Ingestion points: Files read in Step 3 and Step 5 of SKILL.md to detect project structure and tooling.
Boundary markers: None identified; the skill interprets the content of manifest files directly.
Capability inventory: The skill uses Bash, Write, and Edit tools and executes git and agent-validate commands as described in SKILL.md.
Sanitization: There is no evidence of sanitization or strict validation of the content of the manifest files before they are parsed for command discovery.
[COMMAND_EXECUTION]: The skill performs shell command execution for repository operations and tool validation. Examples include running git status, git commit, and agent-validate validate in SKILL.md. These operations are essential for the skill's stated purpose but involve the execution of external binaries.

validator-setup