rhdh-konflux-tasks
Pass
Audited by Gen Agent Trust Hub on Sep 2, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch and follow instructions from external MIGRATION.md files hosted on GitHub (e.g., in the konflux-ci organization). This creates a surface for indirect prompt injection where instructions in the external documentation could influence the agent to perform unintended modifications to the repository.
- Ingestion points: Fetches MIGRATION.md and CHANGELOG.md files from external GitHub URLs; reads local .tekton and .tekton-templates YAML files.
- Boundary markers: The skill instructions emphasize following a /mutation-gate and requiring human review before pushing any changes. It also instructs the agent to skip 'no action required' sections to limit the influence of external content.
- Capability inventory: The skill has the ability to modify local files using sed and execute shell scripts (updateDigests.sh, generatePipelineRuns.sh, check-trusted-tasks.sh).
- Sanitization: No specific sanitization or escaping of the external markdown content is mentioned prior to processing.
- [COMMAND_EXECUTION]: The skill executes several local shell scripts and CLI tools including skopeo, jq, yq, and git. These operations are restricted to the local environment and are central to its purpose of automating CI/CD configuration updates.
- [EXTERNAL_DOWNLOADS]: The scripts/check-trusted-tasks.sh script uses skopeo to download metadata and layers from the quay.io/konflux-ci/tekton-catalog/data-acceptable-bundles container image to verify task security pins. Additionally, references/konflux-migration-urls.md provides logic for fetching documentation from raw.githubusercontent.com. Both sources are recognized services and relevant to the skill's specific technical domain.
Audit Metadata