The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted alert data, such as alert names and IP addresses, from external SIEM or EDR sources and interpolates this data into triage reports and API payloads.
Ingestion points: Input variables and command-line arguments in scripts/agent.py and scripts/process.py used to represent alert content.
Boundary markers: The skill does not implement specific delimiters or instructions to prevent the agent from potentially following instructions embedded within the ingested alert data.
Capability inventory: The skill can write files to the local file system (incident_triage_report.json), make network requests to multiple external APIs, and create cases in case management systems.
Sanitization: No explicit validation or sanitization of the input alert data is performed before it is used in documentation or API calls.
[EXTERNAL_DOWNLOADS]: The skill interacts with external platforms for indicator enrichment and response coordination, including VirusTotal, AbuseIPDB, MITRE ATT&CK, and PagerDuty.
These requests target well-known services and are part of the intended incident response workflow.
[COMMAND_EXECUTION]: Documentation and scripts provide examples of executing shell commands (curl) and inline Python code to automate triage steps.
These commands are used for local processing and interaction with legitimate security infrastructure.

triaging-security-incident-with-ir-playbook