render-to-aws
Pass
Audited by Gen Agent Trust Hub on Nov 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONEXTERNAL_DOWNLOADS
Full Analysis
- [Indirect Prompt Injection Surface]: The skill processes untrusted data from Render Blueprints, application source code, and API responses to generate infrastructure-as-code (Terraform) and migration scripts.
- Ingestion points: Untrusted data enters the agent context through capture-render-api.py, capture-render-cli.py, parse-render-blueprint.py, and scan-render-application.py.
- Boundary markers: Present. The skill normalizes all inputs into structured JSON intermediate files (render-resource-inventory.json) using strict schemas before further processing.
- Capability inventory: High. The skill possesses the capability to generate HCL (Terraform), Shell scripts, and HTML reports based on the ingested data.
- Sanitization: Present. The skill employs rigorous sanitization, including regex-based secret scrubbing in classifier.py, closed projections in normalize-render-capture.py that only retain reviewed fields, and post-generation checks like terraform validate, a gitleaks-based secret scan, and a custom policy validator.
- [Controlled Command Execution]: The skill invokes the render CLI to gather infrastructure metadata.
- Evidence: capture-render-cli.py uses subprocess.Popen with shell=False and a strictly constructed argument list based on a reviewed contract.
- Context: The skill utilizes a hardened environment for child processes, stripping unrelated credentials from the environment before execution to prevent accidental exposure.
- [Supply Chain Security]: The skill manages its runtime environment using uv with pinned dependency versions.
- Evidence: SKILL.md specifies execution through uv run --project scripts --frozen, and scripts/pyproject.toml defines exact versions for libraries like jsonschema and PyYAML.
- [Secure Data Handling]: Extensive logic is implemented to ensure that secrets and credentials from the source environment are not leaked into reports or generated artifacts.
- Evidence: capture-render-api.py and classifier.py use comprehensive regex patterns to detect and redact sensitive keys and values (e.g., passwords, tokens, API keys) from API responses and application code metadata.
Audit Metadata