openspec-archive-change

Pass

Audited by Gen Agent Trust Hub on Mar 30, 2026

Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands (mkdir, mv) using a variable <name> (Step 5). While the instructions mandate user selection from a list, any failure to sanitize this input could allow for shell command injection if an attacker can control the change name stored in the system.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface by reading and parsing external data from local files.
  • Ingestion points: Reads tasks.md (Step 3) and delta specification files in openspec/changes/<name>/specs/ (Step 4).
  • Boundary markers: None identified; the skill directly processes the file content to count tasks and analyze diffs.
  • Capability inventory: Performs file system operations (mkdir, mv) and invokes other skills via a subagent (Step 4, 5).
  • Sanitization: No evidence of sanitization or escaping for the content read from the tasks or specification files before it is used to generate summaries or warnings for the user.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 30, 2026, 09:22 PM