meticulous-zero-diff-task
Pass
Audited by Gen Agent Trust Hub on Oct 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes various Meticulous CLI commands and MCP tools, such as 'meticulous agent upload-build' and 'trigger_test_run', to manage the visual testing lifecycle.
- [INDIRECT_PROMPT_INJECTION]: The skill interacts with project-specific configuration files, creating a surface for indirect prompt injection. * Ingestion points: The skill reads repository CI configuration files (.github/workflows/*.yml, .gitlab-ci.yml, and bitbucket-pipelines.yml) to determine build commands and artifact paths. * Boundary markers: No explicit delimiters or safety instructions are provided to isolate the data read from these files from the agent's execution logic. * Capability inventory: The skill can read local files and execute shell commands via the Meticulous CLI or MCP tools based on the parsed data. * Sanitization: The skill does not implement validation or escaping for the values extracted from the configuration files before they are used as arguments in command execution.
Audit Metadata