change-init

Pass

Audited by Gen Agent Trust Hub on Aug 20, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes local setup scripts (scripts/bash/setup-change-init.sh and scripts/powershell/setup-change-init.ps1) to initialize the environment and identify repository paths. It also uses shell commands like git log and git show to mine historical data.
  • [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection (Category 8) as it ingests untrusted content from commit history and external issue trackers. \n
  • Ingestion points: Processes commit subjects and bodies in Phase 2, and fetches issue titles and summaries from GitHub, GitLab, or JIRA in Phase 5. \n
  • Boundary markers: The skill writes generated drafts to a specific directory (.adlc/drafts/chdr/) and explicitly directs the user to perform manual review using the /change-clarify command before promotion. \n
  • Capability inventory: Includes file system writes for draft creation and shell execution for history retrieval. \n
  • Sanitization: The instructions recommend summarizing issue bodies rather than including raw text to reduce the risk of processing malicious instructions embedded in tickets.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 20, 2026, 02:08 PM
Security Audit — agent-trust-hub — change-init