netdata-config-from-requirements

Pass

Audited by Gen Agent Trust Hub on Oct 13, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes untrusted requirements documents and prose to generate functional configurations and scripts. It lacks explicit delimiters to isolate customer-provided data from system instructions.\n
  • Ingestion points: Customer requirements documents, pasted prose, architecture briefs, or attached files (SKILL.md, rules/extract-requirements.md).\n
  • Boundary markers: The instructions do not define specific delimiters (such as XML tags or special markers) to isolate the requirements document text within the prompt context.\n
  • Capability inventory: The agent can write YAML configurations and shell scripts (netdata/claim.sh) as part of the deliverable bundle (rules/deliverable-bundle.md).\n
  • Sanitization: No specific input sanitization is mentioned, though the skill mandates manual verification of the bundle by an account team and the customer before production use.\n- [DYNAMIC_EXECUTION]: The skill generates executable shell scripts (claim.sh) and instructions to run a Python subprocess command for YAML linting. While the linting command uses the secure yaml.safe_load() function, generating executable content from user-provided requirements constitutes a dynamic execution pattern.\n
  • Evidence: SKILL.md (Verification section) describes running a python -c linting command, and rules/deliverable-bundle.md specifies the creation of a netdata/claim.sh script artifact.
Audit Metadata
Risk Level
SAFE
Analyzed
Oct 13, 2026, 12:26 PM