The Agent Skills Directory

[COMMAND_EXECUTION]: The documentation provides multiple curl and python command examples for discovering GraphQL endpoints and testing them for vulnerabilities. These commands are standard for web application security auditing and are targeted at a user-defined placeholder domain.
[EXTERNAL_DOWNLOADS]: The skill references several established open-source security tools and libraries, including graphql-cop, clairvoyance, requests, and gql. These resources are fetched from public registries or reputable GitHub repositories.
[INDIRECT_PROMPT_INJECTION]: The scripts/agent.py script ingests JSON responses from the target GraphQL server. While it does not execute this content, the ingestion of untrusted data from an external API represents an inherent surface for indirect prompt injection if the output is later processed by an agent without sanitization.
Ingestion points: The script reads response bodies using resp.json() in the GraphQLSecurityAgent._query method.
Boundary markers: No explicit boundary markers or delimiters are used for the data returned from the API.
Capability inventory: The script possesses network capabilities via the requests library to send POST queries to a target URL.
Sanitization: The script parses responses as structured JSON but does not sanitize the contents of specific fields (e.g., error messages) before outputting them.

performing-graphql-security-assessment