Issue 01: CrowdStrike buys into AI red teaming, GitBook's AI-agent breach, and 10 new tools
This is a sample issue of the weekly briefing. It covers November 2 to November 9, 2026. Subscribe below to get the next issues by email.
Issue date: . Posted here as a sample on . Facts were checked against the linked sources. When this issue was written, the directory listed 101 tools. It now lists 102.
1. CrowdStrike signs deals for Dreadnode and Fiddler
On Nov 3 at Fal.Con Europe, CrowdStrike announced definitive agreements to acquire Dreadnode and Fiddler AI. Dreadnode does automated AI security testing of models, applications and agents before they ship. Fiddler covers agent observability and will be sold as Falcon Observe. CrowdStrike says Falcon Observe will be generally available in Q4 FY27. Both deals were still pending at our last check. The press releases do not state a price.
The same day, CrowdStrike added an AI kill switch to Falcon Guardian. It cuts off access to chosen AI models and providers at an AI gateway.
Sources: Dreadnode release, Fiddler release, Falcon Guardian post.
2. GitBook: an attack using AI agents reached infrastructure credentials
GitBook says an attacker using AI agents gained access to some of its infrastructure credentials at 19:00 CET on Nov 2. It took the app and authenticated docs sites offline, replaced every credential and invalidated all customer API tokens.
Its Nov 4 post mortem gives more detail. The attacker used a flaw in the SSO login flow to create sessions acting as GitBook staff. A misconfigured GitHub App then led to internal code repositories and, eventually, a Cloudflare API key. That key exposed production secrets with full infrastructure access through a service account.
GitBook says the attack appears aimed at one customer, but it treats all customer data as exposed. It says it found no evidence that any other customer's content was modified.
If you use GitBook: rotate your GitBook API tokens and any third-party tokens or integration secrets connected to it. Re-authenticate the GitBook MCP server if you use it. Then work through the rest of the checklist on the status page.
Source: GitBook security incident status page.
3. Endpoint agent security got crowded
Five of the ten new listings watch what AI agents do on employee machines or across a company. They are CrowdStrike Falcon Guardian, Pluto Security, Sysdig AI Defense, Zscaler AI Endpoint Security and HOL Guard.
- Pluto Security came out of stealth on Oct 13 with $42M in funding.
- Sysdig AI Defense was announced on Nov 4, and its policies start in alert mode.
- HOL Guard is open source (Apache-2.0). It pauses risky coding-agent actions until the developer approves them.
When you evaluate one, ask how it ties each agent action to a person. Also ask whether it can hold an action for approval instead of only alerting.
Compare them: Agent runtime security and posture tools.
4. Upgrade notes
- NVIDIA OpenShell 0.1.7 (Nov 3): audit logs are now OCSF 1.8.0 records only. This breaks anything downstream that still parses the older 1.1 or 1.3 records. Release notes.
- A.I.G v4.6.9 (Nov 5): MCP-Scan and Agent-Scan fixes, a CloudGPU provider preset, and third-party Agent-Scan provider presets removed. The project says its web platform has no authentication, so keep it off public networks. Release notes.
What changed in the directory (Nov 2 to Nov 9)
- Added 10 tools, taking the directory from 91 to 101.
- Pricing updates from vendor pages: Enkrypt AI (free Explore plan, Launch $149/mo, Scale $1,499/mo), Langfuse, Opik and AgentOps.
- Promptfoo: its GitHub README now says it is part of OpenAI.
- Sentra: a $150M strategic investment led by Capital One Ventures, announced Nov 4.
- Other updates: Kong AI Gateway (vendor correction), plus new descriptions for Straiker, Agent Beacon and BeyondTrust.
- Re-checked 59 existing listings against their own sites and repos that week.
- New comparison: The best AI red teaming tools in 2026: 16 compared.
Browse everything in the directory. If you build or use a tool we are missing, email hello@agentsecurityindex.com.