sdlc-execute-plan
Warn
Audited by Gen Agent Trust Hub on Oct 14, 2026
Risk Level: MEDIUMOBFUSCATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [OBFUSCATION]: The skill contains a hidden ASCII control character (Backspace,
\u0008) embedded in the instructions for status updates. This type of hidden character can potentially be used to evade text-based security filters or mislead certain LLM tokenizers.\n - Evidence: Found in the text: "Use only pending, in-progress, completed, and \u0008locked statuses."\n- [INDIRECT_PROMPT_INJECTION]: The skill processes external markdown artifacts to drive repository changes, creating a vulnerability where malicious instructions in those files could influence implemented code.\n
- Ingestion points: The agent reads instructions and context from
sdlc/<work-item>/spec.md,sdlc/<work-item>/plan.md, andsdlc/<work-item>/intent.md.\n - Boundary markers: Absent. The skill does not provide any instructions to use delimiters or to treat the content of these files as untrusted data.\n
- Capability inventory: The skill allows the agent to modify repository source code, write new tests, and update local markdown artifacts.\n
- Sanitization: Absent. There are no instructions to validate, filter, or escape the content read from the SDLC artifacts before implementation.
Audit Metadata