krillinai-pipeline
Pass
Audited by Gen Agent Trust Hub on Oct 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes shell commands using the
$KRILLINAI_CLIenvironment variable. This is used specifically for dry-run validation of pipeline outputs and is consistent with the skill's stated purpose of providing a developer interface for KrillinAI tools. - [INDIRECT_PROMPT_INJECTION]: The skill processes data from an external file, which represents a potential injection surface.
- Ingestion points: Reads
krillinai_manifest.jsonto determine project state (SKILL.md). - Boundary markers: No explicit markers or delimiters are defined for the manifest data content.
- Capability inventory: Executes subprocess commands via
$KRILLINAI_CLI(SKILL.md). - Sanitization: No explicit sanitization or validation of the JSON manifest is mentioned, relying on the CLI's internal handling of the file.
Audit Metadata