ci-cd-pipeline-builder
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted repository files (e.g.,
package.json,pyproject.toml) to generate CI/CD pipeline configurations. An attacker could embed malicious instructions or shell commands within these files to influence the generated YAML output. - Ingestion points:
scripts/stack_detector.pyandscripts/pipeline_generator.pyreadpackage.jsonscripts and other manifest files. - Boundary markers: Absent. The scripts directly interpolate values from JSON/text files into shell command strings.
- Capability inventory: The generated output is often intended to be executed in CI environments. The scripts themselves perform file reads and writes.
- Sanitization: Limited. While it uses
json.loads, the values extracted (like script names) are directly embedded into generated CI shell commands. - [COMMAND_EXECUTION]: The generator script (
scripts/pipeline_generator.py) takes input from external files or stdin to construct shell commands for the generated YAML. If the input source is compromised, it could lead to the generation of malicious CI workflows. - Evidence:
scripts/pipeline_generator.pyuses logic like `for cmd in lint_cmds + test_cmds + build_cmds: lines.append(f" - run: {cmd}")
wherecmd` is sourced from the detected repository scripts.
Audit Metadata