adversarial-audit

Running deep-audit workflow with multi-phase verification and structured output.

One-time or recurring reliability audit of all harness automations; identifies decision-log gaps and backoff issues

Proactive skill routing audit before routing causes user confusion; prevents silent shadowing

Repo-wide sweeps that cannot fit in one conversation, with plausible-but-wrong findings filtered by adversarial skeptics

Running a complete, adversarially-verified code + app quality audit without making edits; pair with quality-remediate workflow to apply fixes.

Best for automating complex multi-phase fanout-audit processes at scale.

Parallel read-only audits against design claims with completeness criticism.

When you want instant, proof-backed security findings in your IDE or CI without noise and prefer deterministic rules to ML models.

Multi-domain code quality auditing at scale across partitions.

Conducting structured code/design reviews with multiple perspectives in parallel.

Finding how an agent behaves unexpectedly (specification-gaming, misalignment, emergent behavior) before production deployment, complementing generic jailbreak scanners.

Fast multi-perspective code audits with adversarial filtering of false positives; HIGH findings must be independently verified before inclusion in report.

Orchestrate multi-phase security-review-mentions tasks with structured verification and recovery.

Model update audits where capability changes must be verified against confirmed facts.

When you need to verify that tests actually test something and aren't vacuous before committing to green.

Drift detection when you maintain parallel implementations (deprecated + new, A/B experiment, etc).

Catching error-suppression in automated fixes via voting.

Finding security vulnerabilities faster than manual code review.

Multi-dimensional acceptance criteria requiring independent per-AC verification.

Finding logic bugs that escape normal code review by generating adversarial test cases.

High-stakes architectural or strategic decisions requiring adversarial pressure-testing.

Finding subtle bugs or policy violations across large codebases that require consistent multi-agent consensus.

Checking specification conformance against a canonical contract with phrase-based verification.

Vicious pre-commit review of high-stakes specs to catch cross-vendor blind spots

Audit a compliance-critical ERP surface when dimension coverage and adversarial verification are essential.

When you need adversarial security assessment of auth, routing, input handling, and data flow in a production codebase.

multi-phase agent orchestration

accessibility/compliance audits

accessibility/compliance audits

Security review before production launch or after major refactoring

Conducting structured code/design reviews with multiple perspectives in parallel.

Scanning a subset of files for a concrete concern (e.g., hardcoded secrets, missing validation).

Security audit: find and classify variants of a known vulnerability pattern across the entire repo, separating confirmed/probable/false-positive.

When you need multi-agent security review that catches adversarial thinking across code, infra, and supply chain.

security and upgrade planning

Conducting structured code/design reviews with multiple perspectives in parallel.

Finding hidden security vulnerabilities and data integrity bugs in adversarial code review mode.

Audit entire codebase for untested code and cyclomatic-complexity violations.

accessibility/compliance audits