The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from external API endpoints and local JSON files to perform its analysis. This data ingestion represents a surface for indirect prompt injection, as the agent processes potentially attacker-controlled content.
Ingestion points: Untrusted data enters the agent context through requests.get (scripts/agent.py) and json.load (scripts/agent.py).
Boundary markers: Absent. There are no explicit instructions or delimiters used to separate untrusted data from the agent's internal instructions.
Capability inventory: The associated script scripts/agent.py can perform network operations via requests.get and file-write operations via json.dump.
Sanitization: Absent. The skill identifies patterns but does not perform sanitization, escaping, or validation on the data it processes.
[SAFE]: The script scripts/agent.py explicitly disables SSL certificate verification (verify=False). This is common in security testing tools to allow for proxying traffic but is a violation of transport security best practices.

exploiting-excessive-data-exposure-in-api