Autonomous White-Box Exploitation.
Zero False Positives.
Nulltrap deploys specialist offensive agent swarms across your repositories to map multi-file trust boundaries, prove exploitability inside air-gapped microVMs, and open pre-tested fix PRs. If it cannot execute a deterministic PoC, your team never gets paged.
Full-Spectrum White-Box Coverage Across Languages & Attack Surfaces
Show, Don't Warn
Every Finding Ships With a Working Exploit & Pre-Tested Patch
Legacy SAST flags 400 theoretical taint warnings and misses multi-file state bugs. Nulltrap suppresses every hypothesis until its air-gapped verifier proves impact and confirms the patch closes the vector.
[SOURCE] POST /v1/webhooks/ingest
└─ gateway/hmac_verify.go:84
// Verifies HMAC over raw body;
// Go json.Unmarshal keeps LAST key
└─ queue/enqueuer.ts:49
// Streaming parser extracts FIRST
// "org_id" key into job metadata
[SINK] worker/payout_dispatch.ts:192
└─ Executes settlement against victim
org_id with attacker-signed payload [nulltrap-vm:net=none] $ pytest repro_desync.py
Seeded Org A (victim): bal=$250,000
Seeded Org B (attacker): key=whsec_test_b
Sending signed payload with duplicate key:
{"org_id":"org_victim","amt":50000,"org_id":"org_attacker"}
[!] HMAC check passed (bound to org_attacker)
[!] Worker debited org_victim (-$50,000)
[POC_EXIT_0] Cross-tenant mutation proven. --- a/queue/enqueuer.ts
+++ b/queue/enqueuer.ts
- const targetOrg = rawExtractFirstKey(body, "org_id");
+ // Bind exclusively to authenticated principal
+ rejectDuplicateJsonKeys(body);
+ const targetOrg = verifiedPrincipal.orgId;
────────────────────────────────────
Unpatched commit : EXPLOITED
Patched commit : BLOCKED (400 StrictJSON)
Existing suite : 312 / 312 PASS Two-Stage Architecture
Deep Full-Repo Context First. Continuous Diff Guard Second.
Reviewing pull requests in isolation misses 80% of critical bugs because a 30-line diff looks harmless without knowing the auth middleware or state lock 15 files away. Nulltrap fuses both stages into one system.
Baseline Deep Hunt
Specialist mapper agents build a structural Attack-Surface & Invariant Graph across your entire repository—indexing entrypoints, privilege checks, parsers, ORM/SQL sinks, concurrency locks, and outbound trust boundaries.
- → Parallel Principal Investigator (PI) agents synthesize custom fuzzers & exploit scripts
- → Air-gapped sandbox execution filters out 100% of non-exploitable noise
- → Delivered natively as private GitHub Security Advisories + ready-to-merge Fix PRs
Continuous Delta Guard
Once your repository's Attack-Surface Graph is mapped, Nulltrap monitors incoming pull requests and release branches—evaluating every diff against the full architectural graph rather than reading diffs in a vacuum.
- → Graph-aware router wakes deep offensive verification when diffs touch trust boundaries
- → Catches subtle regressions in AI-assisted and high-velocity code before merge
- → Scheduled full-codebase sweeps as frontier reasoning models & exploit classes advance
Dedicated hardware-isolated VM per run. Source mounts into encrypted RAM (`tmpfs`) and self-destructs on completion.
External network egress is severed prior to exploitation. Agents never touch live production infrastructure.
Strict enterprise zero-data-retention inference endpoints. Your proprietary code is never stored or trained on.
Counter-signed Mutual NDA on intake. Pure engineering artifacts delivered directly to GitHub—zero sales calls.
Transparent USD Pricing
No Sales Theater. Pay for Verified Signal.
Start with a risk-free 48-hour Baseline Deep Hunt on your core repository, or lock in continuous autonomous coverage across releases.
100% credited if you upgrade to Continuous Delta Guard within 14 days of delivery.
- ✓ Full-repository Attack-Surface & Invariant Graph mapping
- ✓ Unconstrained multi-agent offensive sweep (auth, logic, races, parsers, SSRF/RCE)
- ✓ Every reported issue includes an executable sandbox PoC + pre-tested Fix PR
- ✓ Zero-Risk Guarantee: If zero valid High/Medium vulnerabilities are uncovered, pay $0
Includes the initial 48h Baseline Deep Hunt at zero setup fee.
- ✓ Everything in 48h Baseline Deep Hunt on onboarding
- ✓ Continuous graph-aware exploitation on security-relevant PRs & release branches
- ✓ Monthly full-codebase sweeps as new frontier models & exploit playbooks ship
- ✓ Direct async engineering channel (Slack / GitHub) + compliance attestation letter
Engineering & Advisories
Latest From /research
Zero-Call Async Intake
Initiate an Autonomous Run
No discovery calls required. Submit your target scope below—we reply asynchronously via email with a counter-signed Mutual NDA, isolated GitHub read-access instructions, and invoice link.