Block unsafe AI and API changes before they merge.

Signed allow, block, and needs-approval decisions your CI can verify.

ShakerScan runs DAST and AI Gate checks against preview deployments, AI APIs, chat widgets, RAG apps, and agent workflows. Each gate returns evidence, policy results, and a signed decision bound to the repo, commit, target, environment, and policy.

Unsafe demo -> BLOCK · Safe demo -> ALLOW · No card required.Create a workspace to run the demo.
GitHub PR gatesSigned evidenceRaw/redacted artifactsScoped approval tokensAPI/widget/RAG/MCP targets

Built for B2B SaaS teams shipping AI APIs, chat widgets, RAG apps, and agents through GitHub/CI.

GitHub check: Shaker Security Gate
BLOCKED
Merge stops
Policy threshold crossed

Finding: RAG cross-tenant retrieval. Evidence hash: sha256:8b7c...e21f.

Repo
acme/support-bot
Commit
9f31c2a
Target
preview-api
Environment
preview
shakerscan gate verify --evidence-id eval_123 --repo acme/support-bot --branch feature/rag-tenant-filters --commit-sha 9f31c2a --environment preview --target-id preview-api --probe-pack shaker-ai-smoke --decision block --evidence-hash sha256:8b7c...e21f
allow

Signed evidence ready. Attestation verified.

block

Policy threshold crossed. Merge stops.

needs_approval

Human exception required for eligible workflows.

Gate surfacesCI/CDAI APIsChat widgetsRAG appsAgent workflowsMCP traces
How It Works

Six steps from pull request to deploy decision.

1

Open PR

A pull request or preview deployment creates the release scope Shaker will test.

2

Test Surface

DAST and AI Gate checks run against the deployed web, API, widget, RAG, or agent surface.

3

Evaluate Policy

Findings are reduced to an allow, block, or needs-approval decision with policy context.

4

Sign Evidence

Evidence, scope, policy, probe pack, and decision are bound into a verifiable artifact.

5

Verify in CI

GitHub or CI verifies the signed decision before the risky change can ship.

6

Ship or Stop

The release proceeds, blocks, or waits for an eligible scoped approval-token workflow.

GitHub Check

See the check developers actually have to satisfy.

ShakerScan publishes a stable required check so branch protection can enforce the signed decision. If the gate returns block or the attestation does not verify, the merge stops.

Shaker Security Gate
Failed
Required

1 high-risk AI finding crossed the release policy threshold. CI can verify the signed evidence before deciding whether this commit is allowed to deploy.

Finding
RAG cross-tenant retrieval
Decision
BLOCK
Evidence
sha256:8b7c...e21f
Attestation
verified
Required check
Shaker Security Gate
shakerscan gate verify --evidence-id eval_123 --repo acme/support-bot --branch feature/rag-tenant-filters --commit-sha 9f31c2a --environment preview --target-id preview-api --probe-pack shaker-ai-smoke --decision block --evidence-hash sha256:8b7c...e21f
Capabilities

Three controls. One deploy decision.

Three gates feed the same policy, evidence, attestation, and CI verification layer.

DAST Checks

Preview deployments, APIs, headers, sessions, exposure, and active checks.

AI Behavior Checks

Chat APIs, widgets, RAG apps, agent traces, and MCP traces.

Approval Controls

Signed evidence, needs-approval workflow, eligible scoped tokens, and revocation.

Evidence Packet

Proof a release system can verify.

Each gate result records the tested scope, policy, decision, evidence hash, and attestation state. That turns a scan result into a release-control artifact.

Repo
acme/support-bot
Commit
9f31c2a
Branch
feature/rag-guardrails
Environment
preview
Target
preview-api
Probe pack
shaker-ai-smoke
Policy
release-strict@2026.04
Decision
block
Evidence hash
sha256:8b7c...e21f
Attestation
verified
Approval token
not issued
Buyer Evidence

Answer AI security questions with evidence, not screenshots.

B2B SaaS buyers increasingly ask how AI features are tested before release. ShakerScan gives product security, AppSec, and platform teams a packet they can reuse for customer reviews and internal approvals.

What AI surface was tested?

Which commit and environment?

Which probe pack and policy?

What was the decision?

Where is the evidence hash?

Who approved the exception?

AI Gate

Verifiable security gates for AI-enabled applications.

Test whether your AI features are safe enough to merge, deploy, or require review. AI Gate extends the same scan, verify, and gate loop to chat APIs, RAG apps, agent workflows, and MCP traces, with cryptographic proof when attestation signing is configured.

Packaged AI Gate Coverage

AI Smokefastest
prompt injection,system prompt leakage,sensitive data leakage
OWASP LLMbroader coverage
LLM01 prompt injection,LLM02 sensitive disclosure,LLM05 output handling
Agent Abuseagent workflows
approval bypass,unauthorized actions,account-boundary abuse
MCPMCP workflows
untrusted MCP server,tool metadata manipulation,overbroad trust
RAG Litegrounded apps
retrieval leakage,citation integrity,cross-tenant retrieval

Configurable Judge System

1
Deterministic
Canary leakage, regex secrets, prompt markers, output violations
2
Regex Classifier
PII patterns, XSS detection, injection policy checks, excessive agency
3
LLM Rubric
Model-evaluated severity, exploitability, and business impact
4
Semantic Policy
Cross-transcript behavioral analysis for attack success detection
OWASP LLM 2025MITRE ATLASOWASP MCP Top 10DSSE-style AttestationsEd25519 Signing Support
Release Control

Use AI testing tools to investigate. Use ShakerScan to enforce.

AI pentesting tools help teams explore behavior. ShakerScan turns deployed-surface test results into signed decisions that GitHub, CI, and approval workflows can act on before the release moves forward.

Investigation tools

Explore prompts, browser flows, agent behavior, findings, and possible impact.

ShakerScan gate

Require a signed allow, block, or needs-approval decision before merge or deploy.

Self-serve path

Run a public DAST preview
Try unsafe and safe AI Gate demos
Install the GitHub App
Map a repo to a target
Require Shaker Security Gate
Verify signed evidence in CI
90-Second Demo

See the deploy gate block, verify, and pass.

The homepage video slot is reserved for the shortest proof path: PR opens, AI Gate blocks a risky change, evidence verifies, the fix lands, and deploy proceeds.

Video script
PR to verified deploy gate
1Developer opens a PR and the preview deployment is available.
2Shaker Security Gate starts from GitHub or CI.
3AI Gate finds RAG cross-tenant retrieval and returns BLOCK.
4Evidence packet shows repo, commit, target, policy, hash, and attestation.
5Developer fixes the issue, the gate passes, CI verifies the signed decision, and deploy proceeds.
Integrations

Fits into your existing workflow.

Choose the surface your workflow already lives in. The decision loop stays the same.

GitHub App

Install once, map repos to targets, and publish the stable Shaker Security Gate check.

CLI

One gate command for CI/CD or local use. No dashboard required.

REST API

20+ endpoints for scans, findings, verification, policy, evidence, and approvals.

MCP Server

AI tools for Claude Code, Cursor, and any MCP client.

Webhooks

Signed events for policy evaluation, evidence creation, and state changes.

Verifiers

CLI, GitHub Action, OPA/Rego, and Kyverno verification examples.

Pricing

Start free, scale with your security needs.

$0

Free Preview

Prove the workflow before connecting production systems.

$49/month

Dev

For one developer adding a basic security gate to one repo.

Recommended
$299/month

Build Gate

For teams that need preview and deploy gates, not just vulnerability reports.

$999/month

Agent Control

For teams that need deploy approvals, agent approvals, and AI security evidence.

See all plans and details14-day Build Gate trial included

Get started in 5 minutes.

Run a public DAST preview, try the AI Gate demo targets, and install the GitHub App from the same workspace.