change-clarify
Pass
Audited by Gen Agent Trust Hub on Nov 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to process and transform markdown files (
ChDR-*.md) containing inferred decisions and context. Maliciously crafted content in these repository files could potentially influence agent behavior or cause context-poisoning during the review. - Ingestion points: The skill reads draft records from the
{REPO_ROOT}/.adlc/drafts/chdr/directory. - Boundary markers: While no formal delimiters are used, the skill includes a 'provenance circuit breaker' instruction, requiring that every decision claim cite a commit SHA or URL to prevent the agent from inventing or obeying unverified claims.
- Capability inventory: The agent has the capability to read and update local files and execute specific project scripts (
scripts/bash/setup-change-clarify.sh). - Sanitization: The instructions explicitly require that issue summaries be paraphrased rather than pasted verbatim to ensure accuracy and prevent potential data leaks from the source files.
- [COMMAND_EXECUTION]: The skill executes local setup scripts (
scripts/bash/setup-change-clarify.shandscripts/powershell/setup-change-clarify.ps1) to initialize the review environment and gather metadata from the repository using Git. These scripts perform standard local filesystem and environment discovery operations.
Audit Metadata