firehose
Warn
Audited by Gen Agent Trust Hub on Oct 29, 2026
Risk Level: MEDIUMCOMMAND_EXECUTIONPRIVILEGE_ESCALATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on a shell script (
scripts/fast-merger.sh) that executes GitHub CLI (gh) commands and an embedded Python script to automate repository management, metadata parsing, and code merging.- [PRIVILEGE_ESCALATION]: Thefast-merger.shscript utilizes the--adminflag with thegh pr mergecommand. This intentionally bypasses GitHub branch protection rules—such as required reviews and status checks—granting the automated process elevated administrative control over the main branch.- [INDIRECT_PROMPT_INJECTION]: The orchestrator is designed to process and act upon untrusted external data from pull requests, which presents a surface for injection attacks. - Ingestion points: Pull request titles and descriptions are fetched from the GitHub API and processed in
scripts/fast-merger.sh, with the orchestrator instructed to review ledger state inSKILL.md. - Boundary markers: The skill does not implement delimiters or specific instructions to isolate untrusted pull request content from the agent's core instructions.
- Capability inventory: The skill has the capability to perform administrative merges to the repository's main branch, execute shell commands, and modify repository state using the GitHub CLI.
- Sanitization: Content is checked against a regular expression for AI attribution strings, but no further sanitization or validation is performed on the pull request text to prevent embedded instructions from influencing the orchestrator.
Audit Metadata