fmway-po
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill uses shell commands to automate git operations including branch creation and worktree management.
- Evidence:
git switch develop && git pull --ff-only && git switch -c po/{slug}inSKILL.md. - Evidence:
git fetch origin develop && mkdir -p .worktrees && git worktree add .worktrees/po-{slug} -b po/{slug} origin/developinSKILL.md. - [INDIRECT_PROMPT_INJECTION]: The skill ingests raw customer needs and existing repository content to produce development tickets and branch identifiers, creating a surface for potential injection.
- Ingestion points: Processes raw user-provided "customer requests", "needs", and internal project documentation such as
PRD.mdandCLAUDE.md. - Boundary markers: No explicit instruction delimiters or isolation warnings are used to prevent the agent from obeying instructions embedded in the ingested data.
- Capability inventory: Executes git commands via the shell and writes files to the local file system (
docs/requirements/{slug}/TICKET.md). - Sanitization: No validation or sanitization of the input data is performed before it is used to generate filenames or branch names (via the
{slug}variable).
Audit Metadata