How much does RunSybil cost?

No public prices, plans or trial; sold through sales after a demo request.

What are the best alternatives to RunSybil?

Based on documented capability coverage across the Agentic Index 14 point taxonomy, the closest alternatives to RunSybil among security and SOC agents are Equixly, Intezer, Repello AI, Arctic Wolf. Each is graded on the same public evidence, and no vendor pays for placement.

Does RunSybil support MCP and tool calling?

Full under the Agentic Index verification standard. RunSybil publishes public evidence of Model Context Protocol (MCP) support, tool calling, and API or SDK extensibility for connecting agents to external tools. Evidence basis: RunSybil publishes a public API, authorized with org scoped API keys, that creates and configures applications, scope, accounts and source code, triggers tests, checks status, pulls reports, queries findings and starts retests. An MCP serve. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does RunSybil give you tracing and observability into agent runs?

Full under the Agentic Index verification standard. RunSybil publishes public evidence of tracing, observability and audit logging into agent runs, so a team can see what an agent did and why. Evidence basis: Every action Sybil takes is logged and visualized. Each finding carries the HTTP request and response pair, reproduction steps, a step by step validation log and the full exploit chain, and each run keeps a history of created, skipped, fail. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Does RunSybil have SOC 2, SSO and role based access control?

Partial under the Agentic Index verification standard. RunSybil publishes some public evidence of security and identity controls, but narrower than the full set of SOC 2, single sign on (SSO) and role based access control (RBAC). Evidence basis: RunSybil holds SOC 2 Type 2, with the report available for review in its trust center, and keeps internal access control and AI acceptable use policies. Customers manage no access controls of their own. API keys are org scoped, and WorkOS S. Graded from public evidence only against the Agentic Index 14 point capability taxonomy. No vendor pays for placement.

Back to vendors

RunSybil

Visit site

Security / SOC agentindependentVerified 2026-09-28

Entry priceContact salesFull pricing detail

AI offensive security platform whose agent system, Sybil, pentests live applications and infrastructure on every pull request, on a set cadence or on demand, chaining findings into exploit paths and validating exploitability before reporting. Pushes findings to GitHub and Linear, with a public API and an MCP server. Co-founded by Ari Herbert-Voss (first security hire at OpenAI) and Vlad Ionescu (formerly offensive security tech lead at Meta), it raised $40M in a round led by Khosla Ventures.

RunSybil builds Sybil, an AI offensive security platform that pentests live applications and infrastructure from a black box starting point, with credentials, OpenAPI specs and source code as optional inputs. A hierarchy of specialized agents tests hypotheses across the application, each owning a domain such as authentication, business logic or injection. Findings are chained into exploit paths, a false positive triage agent screens them, and a validator agent reruns attacks after remediation, with retests stated to complete in under an hour.

Tests run manually, on every pull request or push through GitHub or GitLab (reading the diff and testing what changed), or on a daily, weekly or monthly cadence. Each finding carries the HTTP request and response, reproduction steps, a step by step validation log, the full exploit chain, a CVSS 3.1 score and CWE ID, and every action Sybil takes is logged and visualized. Findings export as PDF, CSV or JSON, push to GitHub and Linear as tickets, and are reachable through a public API with org scoped API keys and an MCP server for coding assistants such as Codex, Claude Code and Cursor.

Customers define scope and what the program protects, and rules of engagement are confirmed before an engagement begins. RunSybil holds SOC 2 Type 2, hosts on AWS and uses OpenAI, Anthropic and OpenRouter as subprocessors, and WorkOS SSO is coming soon. Its uses span autonomous pentesting, adversarial exposure validation, continuous testing for application security and SOC 2 penetration testing. The team works from hubs in San Francisco and New York City.

Vendor details

Canonical URL

https://www.runsybil.com

Funding status

Independent company with hubs in San Francisco and New York City, co-founded by CEO Ari Herbert-Voss and CTO Vlad Ionescu; raised $40M in a round led by Khosla Ventures (homepage announcement).

Company status

independent

Use cases & customers

Primary use cases

continuous autonomous black-box penetration testing on every deploymentcross-layer vulnerability chaining from application to infrastructure compromiseCTEM Phase 4 validation: proving whether findings are actually exploitableper-pull-request security feedback in CI/CDreplacing bug bounties and point-in-time pentests with continuous coverage

Target customers

security-conscious enterprises with critical customer-data systemsfinancial institutions and Fortune 500 companiesengineering teams embedding continuous testing in CI/CDorganizations replacing annual pentests and bug bounties

Deployment options

SaaS (Amazon Web Services listed as a subprocessor); no region or customer-environment option documented

Integrations

GitHub and GitLab connect for testing triggered by pull requests and pushes. Findings push to GitHub and Linear as tickets, mapped one to one and updated incrementally, and Jira support is coming. A public API with org scoped API keys creates and configures applications, scope, accounts and source code, triggers tests, pulls reports, queries findings and starts retests. An MCP server supports most API workflows except launching tests, for assistants such as Codex, Claude Code and Cursor. Findings export as PDF, CSV and JSON.

In practice

Your engineers ship changes every day. Sybil tests on every pull request or push, reads the diff to test what changed and files findings as tickets in GitHub or Linear.

Your team wants findings it can act on, not false alarms. Sybil chains findings into exploit paths, screens them with a false positive triage agent and attaches the request, response and reproduction steps to each one.

Your developers fixed a vulnerability and need proof it is closed. Sybil's validator agent reruns the attack after remediation, and the retest completes in under an hour.

Sources & related URLs

Research sources

https://www.runsybil.com/ https://www.runsybil.com/blog/runsybil-raises-40m-to-build-the-ai-native-platform-for-offensive-security

Agentic Index coverage score

8.0 / 14 capabilities · 57%

Integrations & Tool Calling	Full
Sybil connects to GitHub or GitLab to test on every pull request or push. It pushes findings to GitHub and Linear as tickets, mapped one to one and updated incrementally, and Jira support is coming. The agents act on the customer's live applications and infrastructure during testing. Sourcerunsybil.com/continuous-testing-for-application-securityread 2026-09-28
Workflow Orchestration	Full
A hierarchy of specialized agents tests hypotheses across the application, each owning a domain such as authentication, business logic or injection. Findings are chained into exploit paths, a dedicated false positive triage agent screens them, and a validator agent reruns attacks after remediation. Sourcerunsybil.com/platformread 2026-09-28
Knowledge Grounding & RAG	Partial
Credentials, OpenAPI specs and source code are optional inputs to a black box test, and pull request runs read the diff to test what changed. There is no maintained retrieval structure over the customer's own knowledge. Notion, Confluence and Google Docs integrations are under evaluation. Sourcerunsybil.com/autonomous-pentestingread 2026-09-28
Human Oversight & Guardrails	Partial
The customer defines what the program protects and how success is measured, and rules of engagement are confirmed before an engagement begins. There is no approval step before an exploit or other action goes through. Sourcerunsybil.com/agentic-adversarial-exposure-validation-aevread 2026-09-28
Security, Identity & Governance	Partial
RunSybil holds SOC 2 Type 2, with the report available for review in its trust center, and keeps internal access control and AI acceptable use policies. Customers manage no access controls of their own. API keys are org scoped, and WorkOS SSO is coming soon. Sourcetrust.runsybil.comread 2026-09-28
Observability & Auditability	Full
Every action Sybil takes is logged and visualized. Each finding carries the HTTP request and response pair, reproduction steps, a step by step validation log and the full exploit chain, and each run keeps a history of created, skipped, failed and pending items. Sourcerunsybil.com/platformread 2026-09-28
Memory & State Persistence	Partial
Every engagement updates the same coverage picture of the customer's estate, which persists across runs. RunSybil gives no lifetime for this picture and does not list what it holds. Sourcerunsybil.com/autonomous-pentestingread 2026-09-28
Deployment & Data Residency	Not documented
Amazon Web Services is a subprocessor for RunSybil's own infrastructure. Customers cannot choose a region or run Sybil in their own environment. Tests start from a black box, without source code, and RunSybil does not say where tests run or where findings are stored. Sourcetrust.runsybil.comread 2026-09-28
Prebuilt Agents, Templates & Packs	Partial
Sybil's named agents work inside the product, and the customer does not choose among them. They are the Discovery Agent, the Attack Agent, domain agents for authentication, business logic and injection, a false positive triage agent and a validator agent. Autonomous pentesting, adversarial exposure validation, continuous testing and SOC 2 penetration testing are uses of one product. Tests can be cloned, but there is no template library. Sourcerunsybil.com/platformread 2026-09-28
Triggers & Channel Coverage	Full
Sybil tests on every pull request or push, reading the diff and testing what changed. It also runs on a recurring daily, weekly or monthly cadence the customer defines, and manual runs and API triggered tests are available too. Sourcerunsybil.com/continuous-testing-for-application-securityread 2026-09-28
Model Flexibility & Routing	Partial
OpenAI, Anthropic and OpenRouter are RunSybil's LLM subprocessors, chosen by RunSybil. Customers cannot choose the model. Sourcetrust.runsybil.comread 2026-09-28
APIs, SDKs & MCP Extensibility	Full
RunSybil publishes a public API, authorized with org scoped API keys, that creates and configures applications, scope, accounts and source code, triggers tests, checks status, pulls reports, queries findings and starts retests. An MCP server supports most API workflows except launching tests, for assistants such as Codex, Claude Code and Cursor. RunSybil publishes what the API covers and how it authenticates, but it has no public API reference, and the MCP server's endpoint and tool list are not public. Sourcerunsybil.com/blog/changelog-september-2026read 2026-09-28
Testing, Debugging & Optimization	Not documented
Sybil does not test, evaluate or red team the customer's AI applications or agents. Its validator agent reruns attacks after remediation, and retests complete in under an hour, but those retests check the customer's applications and infrastructure. Sourcerunsybil.com/platformread 2026-09-28
Browser & Computer Use	Not documented
Sybil does not use a browser or a computer. It executes HTTP requests against the target, and findings show each request and response pair. Sourcerunsybil.com/platformread 2026-09-28

The Agentic Index coverage score grades every vendor Full, Partial or Not documented against the same 14 buyer facing capabilities, from public evidence only. Each capability links to how all vendors in the index score on it. How this evidence is graded

Product updates

October 14, 2026·IntegrationsVerified

RunSybil's October release pushes Sybil's pentest findings into Jira as issues with custom field mapping, and updates the existing ticket when a later test rediscovers a finding or a retest completes. Teams can also connect Notion so Sybil's pentesting agents draw application context from the team's knowledge base during a test.

Bears on: Memory / state

View source

September 15, 2026·IntegrationsVerified

RunSybil's September release adds ticketing integrations with Linear and GitHub that push and track security findings automatically, and a public API for configuring applications.

Bears on: Integrations

View source

View all 2 RunSybil updates →Tracked since Sep 2026 · Verified from public vendor sources

Pricing

Contact sales

Inference, not stated by the vendor: the applications and infrastructure in scope for testing.

What is public

Triggers (pull request, push, schedule), GitHub and Linear ticketing, the public API and MCP server, and SOC 2 Type 2 are public; no plans, tiers, prices or trial are published.

Variable cost rationale

Inference, not stated by the vendor: a subscription over the applications in scope is the likeliest shape; no per-finding, per-test or per-retest meter is published.

Sales call required

Yes, required for paid access

Free / trial

Not published

Missing data

Plan structure, billing unit and whether pull request runs or retests are metered are not published.

Agentic Index verified 2026-09-28

Related vendors

Alternatives to RunSybil

The closest documented capability profiles to RunSybil among security and SOC agents tracked by Agentic Index, ordered by similarity on the same 14 point evidence the rankings use. No vendor pays for placement.

Equixly6.0 / 14A lighter documented profile than RunSybil
Intezer7.0 / 14Fuller documented coverage on Human Oversight & Guardrails
Repello AI8.0 / 14Adds documented Testing, Debugging & Optimization
Arctic Wolf9.5 / 14Adds documented Deployment & Data Residency and Testing, Debugging & Optimization
D3 Security (Morpheus)10.5 / 14Adds documented Deployment & Data Residency
depthfirst10.5 / 14Adds documented Testing, Debugging & Optimization and Browser & Computer Use

Similarity is computed from each vendor's Agentic Index coverage score evidence, axis by axis, not from the totals. How this evidence is graded

Compare RunSybil with other vendors →Compare all security and SOC agents →See where RunSybil ranks →

Data confidence: high

Head to head

RunSybil vs XBOW