AI code security tools: the September 2026 ranking
ANSWER FIRST
This is the September 2026 edition of our ranking, kept for reference with its original criteria, weights and totals. It was replaced on 2026-10-11 by the current ranking, which uses six criteria built around coding agents on the developer endpoint.
Which criteria did the September 2026 edition use?
| Criterion | Weight | What it measured |
|---|---|---|
| AI-generated code coverage | 20% | Scanning of code as it is written, at commit and at pull request, whoever or whatever wrote it: SAST, SCA, IaC, and whether findings from other scanners are pulled in. |
| Coding agent guardrails | 30% | Controls that act on the assistant or agent itself: discovery of assistants, models, MCP servers and skills; policy on what agents may run or install; secrets kept out of prompts; guidance given before code is generated; named support for Claude Code, Cursor, GitHub Copilot and others. |
| Secrets and supply chain | 15% | Secrets detection and prevention (in code and beyond it), malicious package blocking, pipeline and SDLC posture, SBOM and AI-BOM. |
| Prioritization and context | 10% | How the tool decides what matters: reachability, exploitability, runtime and business context, de-duplication. |
| Developer workflow fit | 10% | IDE, CLI, PR and MCP presence, and whether a developer can start without a sales process. |
| Deployment effort | 5% | Time and work from sign-up to first useful result, based on published setup material. |
| Pricing transparency | 10% | Whether a buyer can model cost from public pages: published prices score highest, a published pricing unit scores partially, "Contact sales" scores lowest. |
How did the tools rank in September 2026?
| Rank | Tool | AI-generated code coverage (20%) | Coding agent guardrails (30%) | Secrets and supply chain (15%) | Prioritization and context (10%) | Developer workflow fit (10%) | Deployment effort (5%) | Pricing transparency (10%) | Total |
|---|---|---|---|---|---|---|---|---|---|
| 1 | Legit Security | 5.0 | 5.0 | 5.0 | 4.5 | 3.5 | 3.0 | 1.0 | 4.3 / 5 |
| T-2 | Endor Labs | 4.5 | 4.5 | 4.5 | 5.0 | 4.0 | 4.0 | 2.0 | 4.2 / 5 |
| T-2 | Snyk | 4.5 | 4.5 | 3.5 | 3.5 | 4.5 | 4.5 | 3.5 | 4.2 / 5 |
| T-4 | Cycode | 4.5 | 4.5 | 4.5 | 4.5 | 3.5 | 3.0 | 1.5 | 4.0 / 5 |
| T-4 | Semgrep | 4.5 | 3.5 | 4.0 | 4.0 | 4.5 | 4.5 | 4.0 | 4.0 / 5 |
| 6 | OX Security | 4.0 | 4.5 | 4.0 | 4.0 | 3.5 | 3.5 | 2.0 | 3.9 / 5 |
| T-7 | Aikido Security | 4.0 | 2.5 | 4.0 | 3.5 | 4.5 | 5.0 | 5.0 | 3.7 / 5 |
| T-7 | Apiiro | 4.5 | 4.0 | 4.0 | 5.0 | 3.0 | 2.5 | 1.0 | 3.7 / 5 |
| 9 | GitHub Advanced Security | 4.0 | 2.0 | 4.0 | 3.0 | 4.5 | 4.5 | 4.5 | 3.4 / 5 |
What were the reasons behind each September 2026 score?
1. Legit Security, 4.3 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 5.0 | Scans in the IDE through VibeGuard (SAST and SCA), in source control and CI, and pulls findings from existing scanners into one de-duplicated queue. |
| Coding agent guardrails | 5.0 | VibeGuard 2.0 (launched 3 Aug 2026) discovers coding agents on the developer endpoint, blocks dangerous operations, secures MCP use and resists being switched off; Legit names Claude Code, Cursor, GitHub Copilot and Windsurf. |
| Secrets and supply chain | 5.0 | Secrets scanning reaches repos and Git history plus Jira, Confluence, Slack and developers’ personal GitHub accounts; the supply chain module enforces "100’s of policies" across SDLC systems. |
| Prioritization and context | 4.5 | Risk scoring weighs business criticality, compliance, GenAI use, APIs and internet exposure; reachability is not the headline method on the pages reviewed. |
| Developer workflow fit | 3.5 | Reaches developers through the IDE, a CLI and MCP, but there is no free or self-serve tier a developer can try alone. |
| Deployment effort | 3.0 | An enterprise platform sold through demos; value depends on connecting source control, CI/CD and existing scanners. |
| Pricing transparency | 1.0 | No prices, no pricing unit and no free tier are published. |
Update, 13 October 2026: Codex: supported (per Legit). The frozen September reasons list only the assistants named on Legit's public pages at the time.
T-2. Endor Labs, 4.2 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.5 | AI SAST traces dataflow across repositories and pull requests, and AI Security Code Review agents flag design-level issues such as endpoints shipped without authentication. |
| Coding agent guardrails | 4.5 | AURI inventories coding agents, models, MCP servers and skills and applies policy at the moment of action; the MCP server installs into Cursor, Claude Code, VS Code with Copilot, Codex and Antigravity. |
| Secrets and supply chain | 4.5 | Malicious package detection, a package firewall, pre-commit secrets detection, SCA and artifact signing. |
| Prioritization and context | 5.0 | Reachability analysis across dependencies, containers and first-party code is the core of the product. |
| Developer workflow fit | 4.0 | The free Developer tier runs locally with no account. |
| Deployment effort | 4.0 | Local start is one command; policies, UI and scan history need a paid Core or Pro plan. |
| Pricing transparency | 2.0 | Plan names and the seat definition (contributors with commits in the last 90 days) are published; prices are not. |
T-2. Snyk, 4.2 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.5 | SAST, SCA, IaC and container scanning, with real-time scanning in the IDE, CLI and source control. |
| Coding agent guardrails | 4.5 | Snyk Evo Agentic Development Security discovers MCP servers, skills and tools and can steer or block risky agent activity; Snyk names Claude, Cursor, Codex, Windsurf, Antigravity, Kiro and Qodo. |
| Secrets and supply chain | 3.5 | Open source dependency scanning is core and Evo adds MCP and skill discovery; secrets scanning was not described on the pages reviewed. |
| Prioritization and context | 3.5 | The prioritization method is not described on the pages reviewed. |
| Developer workflow fit | 4.5 | The Free plan includes IDE, CLI and source control integrations. |
| Deployment effort | 4.5 | Free and Team plans are self-serve; the agent security features need an Enterprise subscription. |
| Pricing transparency | 3.5 | Free ($0) and Team (starting at $25 a month, up to 10 developers) are published and Enterprise uses a public credit rate card (1 credit = $1), but coding agent security sits in Enterprise only. |
T-4. Cycode, 4.0 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.5 | SAST, SCA, secrets, container and IaC scanning, plus agentic code scanning and an AI Fix and Remediation Agent that produces PR-ready fixes. |
| Coding agent guardrails | 4.5 | IDE guardrails intercept secrets, file context, prompts and tool calls before they reach AI services, and the AI-BOM tracks rule files, MCP servers and AI API keys; Cycode names Copilot, Cursor, Windsurf and Tabnine on its AI page, and its AI Guardrails blog post names Claude Code and Cursor. |
| Secrets and supply chain | 4.5 | CI/CD pipeline posture, source code leakage detection, secrets, SBOM and AI-BOM. |
| Prioritization and context | 4.5 | An AI Exploitability Agent confirms whether a CVE is reachable. |
| Developer workflow fit | 3.5 | IDE and MCP presence and a free trial, but no free tier. |
| Deployment effort | 3.0 | Breadth (100+ integrations through ConnectorX) means a planned rollout. |
| Pricing transparency | 1.5 | Tiers are named and the pricing basis is stated (active developers and AI usage); no figures are published. |
T-4. Semgrep, 4.0 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.5 | Semgrep Code combines AI reasoning with rules (Semgrep Multimodal) and puts remediation guidance in PRs and IDEs. |
| Coding agent guardrails | 3.5 | Semgrep Guardian (MCP server, hooks on every file write, skills) works in Claude Code, Cursor and Windsurf, but the pages reviewed show no org-wide discovery of assistants or MCP servers, and Copilot is not listed. |
| Secrets and supply chain | 4.0 | Separate Supply Chain and Secrets products, and Guardian blocks malicious dependencies before install. |
| Prioritization and context | 4.0 | Semgrep Assistant auto-triages findings and uses code context to cut false positives. |
| Developer workflow fit | 4.5 | Free for up to 10 contributors and 10 private repos, with VS Code and JetBrains support. |
| Deployment effort | 4.5 | Self-serve start on Free and Teams. |
| Pricing transparency | 4.0 | Teams is $30 a month per contributor for Code or Supply Chain and $15 for Secrets; Guardian pricing is not stated. |
6. OX Security, 3.9 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.0 | OX Code covers SAST, SCA, secrets, IaC, containers and API scanning, with pipeline scanning and IDE and CLI tools. |
| Coding agent guardrails | 4.5 | OX VibeSec reads the developer's prompt and sends security guidance to the agent before code is generated, and adds malicious dependency blocking, secret protection and an AI inventory; supports Claude Code, Cursor, Kiro and VS Code with Copilot. |
| Secrets and supply chain | 4.0 | Secrets and PII detection, SBOM with drift detection, and software supply chain security. |
| Prioritization and context | 4.0 | ASPM correlation plus an Agentic Pentester that ties exploitable issues back to code and APIs. |
| Developer workflow fit | 3.5 | VS Code extension, JetBrains plugin and MCP integration. |
| Deployment effort | 3.5 | A "Start Free" sign-up is offered; its limits are not published. |
| Pricing transparency | 2.0 | The pricing unit for each product is published (per AI user, per contributing developer, per cloud asset, per agent hour); prices are not. |
T-7. Aikido Security, 3.7 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.0 | SAST, AI SAST, secrets and SCA on every plan, with PR Security Review from the Basic plan. |
| Coding agent guardrails | 2.5 | The IDE plugin scans for SAST issues and secrets in VS Code, Cursor and Windsurf, but the pages reviewed show no agent inventory, MCP governance or agent policy controls. |
| Secrets and supply chain | 4.0 | Secrets, SCA, license risk, malware detection (Pro) and Aikido Safe Chain, which blocks malware at package install. |
| Prioritization and context | 3.5 | The prioritization method is not described on the pages reviewed. |
| Developer workflow fit | 4.5 | Free IDE plugin and AutoFix pull requests. |
| Deployment effort | 5.0 | A free-forever plan for 2 users. |
| Pricing transparency | 5.0 | Flat prices are published: Developer $0 (2 users), Basic $300 a month and Pro $600 a month (10 users included). |
T-7. Apiiro, 3.7 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.5 | AI SAST, AI SCA, pre-commit prevention and Deep Code Analysis that detects material code changes. |
| Coding agent guardrails | 4.0 | Apiiro Guardian Agent adds secure prompt enrichment and works with Cursor, GitHub Copilot and Claude Code; its MCP server needs an Apiiro account with API access. |
| Secrets and supply chain | 4.0 | Software supply chain security, AI-BOM discovery and secrets security. |
| Prioritization and context | 5.0 | A Software Architecture Graph and Risk Graph connect code to runtime context. |
| Developer workflow fit | 3.0 | Developer surface through MCP and AutoFix; no free tier for commercial teams (AutoFix is free for open source maintainers). |
| Deployment effort | 2.5 | A demo-led enterprise platform built on deep analysis of the codebase. |
| Pricing transparency | 1.0 | No pricing is published. |
9. GitHub Advanced Security, 3.4 / 5
| Criterion | Score | Why |
|---|---|---|
| AI-generated code coverage | 4.0 | CodeQL code scanning, dependency review and Copilot Autofix; the Copilot coding agent also runs CodeQL, Advisory Database and secret scanning checks on its own changes without a GHAS license. |
| Coding agent guardrails | 2.0 | Built-in checks apply to GitHub's own Copilot coding agent; the pages reviewed show no controls for third-party assistants. |
| Secrets and supply chain | 4.0 | Secret scanning with push protection and AI-detected secrets, plus Dependabot alerts. |
| Prioritization and context | 3.0 | Security overview and campaigns; no reachability or business-context scoring on the pages reviewed. |
| Developer workflow fit | 4.5 | Native to pull requests on GitHub. |
| Deployment effort | 4.5 | Switched on per organization on GitHub Team or GitHub Enterprise Cloud, for repositories on GitHub. |
| Pricing transparency | 4.5 | GitHub Secret Protection is $19 and GitHub Code Security is $30 per active committer per month. |