ci-fails-buildkite
Pass
Audited by Gen Agent Trust Hub on Jul 22, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [COMMAND_EXECUTION]: The skill invokes local shell scripts
.buildkite/scripts/ci-fetch-log.shand.buildkite/scripts/ci-clean-log.shto automate log management tasks within the repository context. - [EXTERNAL_DOWNLOADS]: Fetches log data from the official Buildkite domain, which is an industry-standard CI/CD platform.
- [PROMPT_INJECTION]: The skill ingests and processes CI logs, which are external data sources that could potentially contain adversarial content intended to mislead the agent.
- Ingestion points: Log files downloaded via the
.buildkite/scripts/ci-fetch-log.shscript. - Boundary markers: No explicit markers are defined in the skill instructions to delimit log content.
- Capability inventory: Authorization to run repository-local shell scripts and read the resulting log files.
- Sanitization: Log content is pre-processed by the fetch script to remove timestamps and ANSI escape codes before being presented for analysis.
Audit Metadata