The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to ingest external data, including example API responses and documentation, to generate GraphQL schemas and test fixtures. This creates a surface where malicious instructions embedded in the ingested data could attempt to influence the agent's code generation or subsequent tool usage.
Ingestion points: User-provided API responses and external documentation retrieved via MCP tools during the research phase (SKILL.md Step 1).
Boundary markers: The skill does not provide specific delimiters or 'ignore' instructions for the data being processed.
Capability inventory: The agent has permission to write and edit local files and execute rover CLI commands via Bash (SKILL.md frontmatter).
Sanitization: No specific filtering or sanitization steps are defined for external content before its use in schema templates.
[COMMAND_EXECUTION]: The skill uses the official rover CLI to perform core tasks such as supergraph composition, connector execution, and automated testing. These commands are standard for the GraphQL development workflow and are restricted to the rover binary as specified in the platform configuration.
Evidence: Commands such as rover supergraph compose, rover connector run, and rover connector test are defined in the Process section and validation guide (SKILL.md, references/validation.md).
[EXTERNAL_DOWNLOADS]: The skill includes a recommendation for the user to install a successor skill using a CLI command that targets the author's official GitHub organization.
Evidence: The deprecation notice recommends running npx skills add apollographql/graphos-factory (SKILL.md).

apollo-connectors