/

CVE-2026-61436

Severity:

HIGH RISK

(8.6)

Authentication Bypass (CWE-287) in PraisonAI Webhooks

CVE-2026-61436

Deferred

Vulnerability type

Category:

Authentication Bypass

CWE:

CWE-287

Impact:

Unauthorized Agent Invocation and Message Forgery

Severity:

HIGH RISK

(8.6)

CVSS version:

3.1

Catch this CVE while code is being written

Detect vulnerable code directly in pull requests, so CVEs are fixed before merge.

Summary

PraisonAI before 4.6.78 does not verify Svix webhook signatures in AgentMail webhook mode, enabling unauthenticated attackers to forge message.received events.

By sending crafted JSON payloads to the exposed webhook endpoint, an attacker can trigger configured agents with arbitrary sender identities and message content.

Affected context

This vulnerability appears in PraisonAI deployments that use the AgentMail webhook mode and accept incoming Svix-style webhooks over HTTP or HTTPS.

It is relevant where the webhook endpoint is reachable from external networks or untrusted services and relies on Svix signatures for authenticity.

Environments that treat webhook-triggered events as trusted input to downstream agents, workflows, or integrations are particularly exposed.

Exploit preconditions

  • Attack surface: Public or internally accessible AgentMail webhook endpoint
  • Required access: Network-level ability to send HTTP POST requests to the webhook URL
  • Trigger point: Processing of incoming message.received webhook events without validating Svix signatures
  • Why exploitation is possible: The application fails to authenticate webhook requests, accepting any well-formed JSON payload as a legitimate event and invoking configured agents based on forged data

Why this happens

The issue stems from treating webhook payloads as trusted without enforcing robust request authentication.

In the affected AgentMail mode, PraisonAI omits or incorrectly implements Svix signature verification on incoming requests, mapping directly to CWE-287: Improper Authentication.

This insecure pattern allows any actor capable of reaching the endpoint to simulate legitimate event sources, bypassing expected trust boundaries between webhook senders and the application.

Potential impact

  • Attackers can invoke agents with arbitrary sender addresses and fabricated message content
  • Agents may act on forged instructions, leading to unintended data access or actions in integrated systems
  • Workflow automation triggered by message.received events can be abused for spam, phishing, or social engineering campaigns
  • Logging, auditing, and monitoring data can be polluted with fake conversations and events
  • Downstream services relying on PraisonAI output may process malicious or deceptive content

General mitigation guidance

Mitigation generally involves enforcing strict authentication and integrity checks on all webhook traffic and reducing implicit trust in event sources.

Common defensive patterns include:

  • Verifying Svix (or equivalent provider) signatures on every webhook request using shared secrets or public keys
  • Rejecting any webhook request that fails signature validation or originates from unknown sources
  • Limiting webhook endpoint exposure through network controls, such as IP allowlists or private ingress configurations
  • Treating webhook-derived message content as untrusted input and applying validation, sanitization, and policy checks before invoking agents
  • Designing agents and workflows with least-privilege behavior so that forged events have limited impact

How teams detect issues like this

Organizations can detect this class of issue by:

  • Reviewing webhook handling code paths to confirm that signature verification or equivalent authentication is consistently enforced
  • Using static analysis or linters to flag HTTP endpoints that process external JSON payloads without upstream authentication or signature checks
  • Adding security-focused code review checklists for webhook integrations, including verification of headers, signatures, and shared secrets
  • Implementing runtime monitoring and anomaly detection to identify unusual patterns of webhook traffic, such as spikes in message.received events or unexpected sender identities
  • Employing automated application security testing tools to exercise webhook endpoints and verify that unauthenticated requests are rejected

Sources

SourceLink
github.comgithub.com/MervinPraison/PraisonAI/commit/2a855c470077c7d2e2479a575f7ef7f548d51c33
github.comgithub.com/MervinPraison/PraisonAI/commit/846568c7a5d8ce9e71e56e4c213f027c04909753
github.comgithub.com/MervinPraison/PraisonAI/security/advisories/GHSA-7c92-x8vg-4258
vulncheck.comvulncheck.com/advisories/praisonai-before-missing-webhook-signature-verification

Frequently asked

questions

If the AgentMail webhook endpoint is reachable from the internet or any untrusted network segment, any actor with network access can attempt to exploit it by sending crafted JSON payloads.

The attacker only needs the ability to send HTTP POST requests to the webhook URL; no prior authentication or valid Svix credentials are required when signature verification is missing.

Enforce strict webhook request authentication by validating Svix signatures (or equivalent mechanisms) for every request, rejecting any payload that fails verification, and ensure this logic cannot be bypassed.

Treat all inbound webhooks as untrusted by default, require cryptographic verification or strong authentication for every call, document these requirements in integration guidelines, and enforce them with code review policies and automated security checks in CI/CD.

Find out what's already
exploitable in your codebase.