teamcity-first-green-build
Pass
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes local build tools (such as
./gradlewand./mvnw) and validation commands (usingpython3or theteamcityCLI) to inspect the repository and verify TeamCity configurations. - [INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted data from build logs, repository configuration files, and TeamCity API responses while diagnosing failures and setting up pipelines.
- Ingestion points: Build logs, repository files (e.g.,
build.gradle,.teamcity.yml), and TeamCity API outputs (e.g.,problemOccurrences,failed tests). - Boundary markers: The instructions focus on specific debugging orders and workflows but do not define explicit prompt delimiters for interpolated external data.
- Capability inventory: Local command execution for build and validation, file system read access, and authenticated TeamCity API interaction.
- Sanitization: The skill incorporates robust security rules, including mandatory secret redaction and the use of secure YAML parsing (
yaml.safe_load) for configuration files. - [DYNAMIC_EXECUTION]: The skill facilitates the local execution of project-specific configuration code (Kotlin DSL via Maven) and YAML validation scripts to verify TeamCity settings before deployment.
Audit Metadata