fable-heal
Warn
Audited by Gen Agent Trust Hub on Oct 7, 2026
Risk Level: MEDIUMDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [DYNAMIC_EXECUTION]: The skill synthesizes code patches and TDD regression tests (stored in 'test/security/redteam-healed.test.ts') which are subsequently executed using the 'bun test' runner. This represents runtime script generation and execution based on provided vulnerability finding data.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted vulnerability reports ('redteam_findings') to determine remediation strategies, creating a surface for injection.
- Ingestion points: The 'redteam_findings' input, which can be an array or a file path to SARIF/JSON reports.
- Boundary markers: Absent; the skill lacks explicit delimiters or instructions to prevent the agent from following instructions embedded within the findings.
- Capability inventory: The skill possesses extensive capabilities including workspace mutation, file system writes, and shell command execution ('bun test').
- Sanitization: While the skill performs syntax and type verification on generated patches, it does not explicitly sanitize the content of the input findings against malicious instruction injection.
- [COMMAND_EXECUTION]: The skill invokes shell commands via the Bun test runner to execute and verify the generated security regression tests.
Audit Metadata