openspec-archive-change
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill interpolates the
<name>variable directly into shell commands in Step 2 (openspec status --change "<name>") and Step 5 (mv openspec/changes/<name> ...). If the change name contains shell-active characters like semicolons or pipes, it could lead to unintended command execution. - [PROMPT_INJECTION]: The skill is vulnerable to indirect prompt injection in Step 3 when processing external data.
- Ingestion points: Reads and parses the content of the
tasks.mdfile. - Boundary markers: None identified; the agent parses the markdown file directly without delimiters or instruction protection.
- Capability inventory: The agent can execute shell commands (
mkdir,mv,openspec) and write to the file system. - Sanitization: No escaping or validation is performed on the content of
tasks.mdor the interpolated change name before execution.
Audit Metadata