The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructions utilize the $ARGUMENTS placeholder to interpolate user-provided input directly into shell commands such as dig, whois, curl, and nmap. This pattern creates a command injection surface if the executing environment does not sanitize metacharacters, potentially allowing arbitrary command execution.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze data from untrusted external sources, including web searches, web fetches, and tool outputs (e.g., nmap scans, certificate transparency logs). This creates a vulnerability surface where malicious content embedded in the target infrastructure or search results could attempt to influence the agent's behavior.
Ingestion points: External data is brought into the agent's context via WebSearch, WebFetch, and the output of the curl command to crt.sh and nmap scans (SKILL.md).
Boundary markers: The instructions do not define specific delimiters or instructions to ignore embedded commands when processing these external findings.
Capability inventory: The skill has access to Bash for command execution and Write for file system operations, which could be abused if an injection is successful.
Sanitization: There are no instructions provided for sanitizing or filtering the content retrieved from external targets before analysis.
[SAFE]: The skill uses crt.sh, which is a well-known and standard service for querying certificate transparency logs.
[SAFE]: The skill includes explicit instructions and boundaries requiring the agent to confirm authorization and scope before performing any active reconnaissance tasks.

iblai-vibe-security-recon