Agent
Static scanner for financial correctness bugs (BigDecimal misuse, float money, rounding/scale, idempotency, lock/concurrency) in Java/Spring settlement code. Stage 1 DETECT (correctness track).
Agent
Post-mortem knowledge distiller — turns each fixed bug/vuln into reusable detection rules, patterns, and financial invariants, then triggers RECYCLE. Stage 5 GUARD + RECYCLE.
Agent
An agent for improving defect-detection guidance through repeated measurement, analysis, proposed changes, blind testing, and regression checks. A blind test evaluates results without revealing the answers, while human approval is still required before a proposal is adopted.
Agent
Reproduces a bug as a minimal failing test (MRS); for security findings builds an attack PoC that violates a financial invariant (succeeds pre-fix, fails post-fix). Stage 2.5 REPRODUCE.
Agent
Root-cause analysis & fix designer — runs 5-Why to the design level, picks a fix strategy, and writes the fix plan + tests so the Stage 2.5 reproduction flips from red to green. Stage 3 FIX.
Agent
Adversarial taint source→sink scanner for financial security/fraud vulns — IDOR over-withdrawal, amount tampering, TOCTOU double-spend, callback forgery, replay, mass assignment. Stage 1 DETECT (security track).
Agent
Builds money-flow attack hypotheses with STRIDE-FIN before scanning, upgrading detection from feature-driven to hypothesis-driven. Stage 0 THREAT-MODEL.
Agent
Closed-loop fix verifier — replays the reproduction/attack PoC, asserts financial invariants under property/fuzz/shadow traffic, and gates the fix on "attack no longer succeeds ∧ invariants always hold". Stage 4 VERIFY.