The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to run vendor-specific CLI utilities, agent-postmark and mockpostmark, to interact with Postmark's infrastructure for diagnostics and management as defined in SKILL.md and references/commands.md.
[INDIRECT_PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection due to its handling of external email data.
Ingestion points: External data including email bodies and headers are retrieved using agent-postmark messages content and messages search commands found in SKILL.md and references/commands.md.
Boundary markers: The skill advises against routine use of full content dumps and suggests specific investigate commands to triage issues safely, as detailed in references/investigations.md and references/output.md.
Capability inventory: The skill allows for command execution via Bash that can modify delivery states, such as activations and suppressions, based on diagnostic results.
Sanitization: Evidence in references/output.md confirms that the CLI tool performs redaction on sensitive fields like tokens, secrets, and raw email blobs before providing output to the agent, reducing the risk of accidental exposure or manipulation.

agent-postmark