brainstormer
01Agent Claude Code
Broad, fast ideation of possible weaknesses and attacker goals; appends to audit-state/BRAINSTORM.md.
Agent Claude Code
Broad, fast ideation of possible weaknesses and attacker goals; appends to audit-state/BRAINSTORM.md.
Agent Claude Code
Orchestrates the full audit pipeline (brainstorm → local → global → validate → threat model → report).
Agent Claude Code
Cross-component/system-level analysis; records plausible multi-component issues as individual files.
Agent Claude Code
File-scoped security auditor; adds ideas to BRAINSTORM.md and creates individual issue files in issues/plausible/.
Agent Claude Code
Performs quality control: verifies audit is fully complete and identifies mistakes made by other agents. This agent MUST be used for the quality control step of the audit.
Agent Claude Code
Produces final audit report in report/ using docs/example-report as style reference.
Agent Claude Code
An expert security proof writer to help identify vulnerabilities, informed by (failed) attempts to prove security.
Agent Claude Code
For surveying a codebase and generating a list of the audit target's features as input to threat modeling and brainstorming.
Agent Claude Code
Maintains ICTM threat model in audit-state/THREATMODEL.md using user-facing invariants.
Agent Claude Code
Confirms or invalidates specific issues; moves files between plausible, confirmed, and invalid directories.
Command Claude Code
Generate a prioritized summary of audit state for report writing.
Command Claude Code
Run the full coordinator-driven audit workflow.
Command Claude Code
Automatically generate audit instructions and gather relevant context.