write-fix
Pass
Audited by Gen Agent Trust Hub on Jun 13, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it instructs the agent to process and act upon bug reports, which are external and untrusted data sources. A malicious bug report could contain instructions intended to override agent behavior during reproduction or testing.
- Ingestion points: External bug reports referenced in
references/task-template.md(via the{{path}}placeholder). - Boundary markers: Absent; the instructions do not provide delimiters or specific warnings to ignore instructions embedded within the bug reports.
- Capability inventory: Shell command execution for reproducing bugs, running test suites, and executing project-specific commands.
- Sanitization: Absent; there is no requirement to sanitize or validate the content of the reproduction steps provided in the reports.
- [COMMAND_EXECUTION]: The skill relies on executing shell commands resolved from a local project file (
AGENTS.md). While it includes a safety measure—instructing the agent to ask the user if a command is undefined—it still grants the agent the capability to run arbitrary tools and scripts in the workspace context to reproduce defects and verify fixes.
Audit Metadata