The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user data, such as primary business domains and target sending volumes, during the intake phase in SKILL.md. This information is later used in research tasks and tool calls. The instructions do not include explicit boundary markers or sanitization guidelines for this data, which creates a potential surface for prompt injection through user-supplied strings.
Ingestion points: Phase 1: Intake section in SKILL.md.
Boundary markers: None identified in the prompt templates.
Capability inventory: Shell command execution for DNS verification (dig in Step 4E of SKILL.md) and local Python script execution (scripts/check-output.py).
Sanitization: No explicit instructions for escaping or validating user-provided domain names before use.
[COMMAND_EXECUTION]: The skill utilizes a local validation script, scripts/check-output.py, and directs the agent to execute shell commands such as dig for DNS record verification. While the script performs benign string matching, these activities represent the execution of local code and command-line tools based on the skill's instructions.

domain-infrastructure