CVE-2026-56832 (GCVE-0-2026-56832)

Vulnerability from cvelistv5 – Published: 2026-10-23 11:55 – Updated: 2026-10-23 11:55

Title

PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Summary

PraisonAI is a multi-agent teams system. From version 4.5.2 to before version 4.6.59, praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes. As a result, any user who can post in the configured approval channel can approve a pending high-risk tool call by sending yes after the approval message appears. the same PoV also shows that the Slack and Telegram messaging approval backends have no explicit approver allowlist parameter, but the primary report-grade issue is the Discord backend's unthreaded channel cross-talk: the approving message does not need to be a reply or otherwise request-bound. This issue has been patched in version 4.6.59.

Severity

8.8 (High)


                        
                          CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

CWE

CWE-345 - Insufficient Verification of Data Authenticity
CWE-863 - Incorrect Authorization

Assigner

GitHub_M CNA under the mitre root

CNA scorecard C 72/100 over 7761 records in the last 180 days details

References

1 reference

URL	Tags
https://github.com/MervinPraison/PraisonAI/securi…	x_refsource_CONFIRM

Impacted products

1 product

Vendor	Product	Version	CPE status
MervinPraison	PraisonAI	Affected: >= 4.5.2, < 4.6.59	guessed

Show details on NVD website

JSON

To clipboard

{
  "containers": {
    "cna": {
      "affected": [
        {
          "product": "PraisonAI",
          "vendor": "MervinPraison",
          "versions": [
            {
              "status": "affected",
              "version": "\u003e= 4.5.2, \u003c 4.6.59"
            }
          ]
        }
      ],
      "descriptions": [
        {
          "lang": "en",
          "value": "PraisonAI is a multi-agent teams system. From version 4.5.2 to before version 4.6.59, praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes. As a result, any user who can post in the configured approval channel can approve a pending high-risk tool call by sending yes after the approval message appears. the same PoV also shows that the Slack and Telegram messaging approval backends have no explicit approver allowlist parameter, but the primary report-grade issue is the Discord backend\u0027s unthreaded channel cross-talk: the approving message does not need to be a reply or otherwise request-bound. This issue has been patched in version 4.6.59."
        }
      ],
      "metrics": [
        {
          "cvssV3_1": {
            "attackComplexity": "LOW",
            "attackVector": "NETWORK",
            "availabilityImpact": "HIGH",
            "baseScore": 8.8,
            "baseSeverity": "HIGH",
            "confidentialityImpact": "HIGH",
            "integrityImpact": "HIGH",
            "privilegesRequired": "LOW",
            "scope": "UNCHANGED",
            "userInteraction": "NONE",
            "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
            "version": "3.1"
          }
        }
      ],
      "problemTypes": [
        {
          "descriptions": [
            {
              "cweId": "CWE-345",
              "description": "CWE-345: Insufficient Verification of Data Authenticity",
              "lang": "en",
              "type": "CWE"
            }
          ]
        },
        {
          "descriptions": [
            {
              "cweId": "CWE-863",
              "description": "CWE-863: Incorrect Authorization",
              "lang": "en",
              "type": "CWE"
            }
          ]
        }
      ],
      "providerMetadata": {
        "dateUpdated": "2026-10-23T11:55:53.084Z",
        "orgId": "a0819718-46f1-4df5-94e2-005712e83aaa",
        "shortName": "GitHub_M"
      },
      "references": [
        {
          "name": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m",
          "tags": [
            "x_refsource_CONFIRM"
          ],
          "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m"
        }
      ],
      "source": {
        "advisory": "GHSA-8579-rgg5-ph2m",
        "discovery": "UNKNOWN"
      },
      "title": "PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals"
    }
  },
  "cveMetadata": {
    "assignerOrgId": "a0819718-46f1-4df5-94e2-005712e83aaa",
    "assignerShortName": "GitHub_M",
    "cveId": "CVE-2026-56832",
    "datePublished": "2026-10-23T11:55:53.084Z",
    "dateReserved": "2026-06-23T14:55:09.117Z",
    "dateUpdated": "2026-10-23T11:55:53.084Z",
    "state": "PUBLISHED"
  },
  "dataType": "CVE_RECORD",
  "dataVersion": "5.2",
  "vulnerability-lookup:meta": {
    "epss": {
      "cve": "CVE-2026-56832",
      "date": "2026-10-24",
      "epss": "0.00223",
      "percentile": "0.12299"
    },
    "nvd": {
      "cve": {
        "affected": [
          {
            "affectedData": [
              {
                "product": "PraisonAI",
                "vendor": "MervinPraison",
                "versions": [
                  {
                    "status": "affected",
                    "version": "\u003e= 4.5.2, \u003c 4.6.59"
                  }
                ]
              }
            ],
            "source": "security-advisories@github.com"
          }
        ],
        "cveTags": [],
        "descriptions": [
          {
            "lang": "en",
            "value": "PraisonAI is a multi-agent teams system. From version 4.5.2 to before version 4.6.59, praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes. As a result, any user who can post in the configured approval channel can approve a pending high-risk tool call by sending yes after the approval message appears. the same PoV also shows that the Slack and Telegram messaging approval backends have no explicit approver allowlist parameter, but the primary report-grade issue is the Discord backend\u0027s unthreaded channel cross-talk: the approving message does not need to be a reply or otherwise request-bound. This issue has been patched in version 4.6.59."
          }
        ],
        "id": "CVE-2026-56832",
        "lastModified": "2026-10-23T12:18:14.560",
        "metrics": {
          "cvssMetricV31": [
            {
              "cvssData": {
                "attackComplexity": "LOW",
                "attackVector": "NETWORK",
                "availabilityImpact": "HIGH",
                "baseScore": 8.8,
                "baseSeverity": "HIGH",
                "confidentialityImpact": "HIGH",
                "integrityImpact": "HIGH",
                "privilegesRequired": "LOW",
                "scope": "UNCHANGED",
                "userInteraction": "NONE",
                "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
                "version": "3.1"
              },
              "exploitabilityScore": 2.8,
              "impactScore": 5.9,
              "source": "security-advisories@github.com",
              "type": "Secondary"
            }
          ]
        },
        "published": "2026-10-23T12:18:14.437",
        "references": [
          {
            "source": "security-advisories@github.com",
            "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m"
          }
        ],
        "sourceIdentifier": "security-advisories@github.com",
        "vulnStatus": "Deferred",
        "weaknesses": [
          {
            "description": [
              {
                "lang": "en",
                "value": "CWE-345"
              },
              {
                "lang": "en",
                "value": "CWE-863"
              }
            ],
            "source": "security-advisories@github.com",
            "type": "Primary"
          }
        ]
      }
    }
  }
}

FKIE_CVE-2026-56832

Vulnerability from fkie_nvd - Published: 2026-10-23 12:18 - Updated: 2026-10-23 12:18

Severity

8.8 (High) - CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

Summary

References

	URL	Tags
	security-advisories@github.com	https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m

Impacted products

	Vendor	Product	Version

JSON

To clipboard

{
  "affected": [
    {
      "affectedData": [
        {
          "product": "PraisonAI",
          "vendor": "MervinPraison",
          "versions": [
            {
              "status": "affected",
              "version": "\u003e= 4.5.2, \u003c 4.6.59"
            }
          ]
        }
      ],
      "source": "security-advisories@github.com"
    }
  ],
  "cveTags": [],
  "descriptions": [
    {
      "lang": "en",
      "value": "PraisonAI is a multi-agent teams system. From version 4.5.2 to before version 4.6.59, praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes. As a result, any user who can post in the configured approval channel can approve a pending high-risk tool call by sending yes after the approval message appears. the same PoV also shows that the Slack and Telegram messaging approval backends have no explicit approver allowlist parameter, but the primary report-grade issue is the Discord backend\u0027s unthreaded channel cross-talk: the approving message does not need to be a reply or otherwise request-bound. This issue has been patched in version 4.6.59."
    }
  ],
  "id": "CVE-2026-56832",
  "lastModified": "2026-10-23T12:18:14.560",
  "metrics": {
    "cvssMetricV31": [
      {
        "cvssData": {
          "attackComplexity": "LOW",
          "attackVector": "NETWORK",
          "availabilityImpact": "HIGH",
          "baseScore": 8.8,
          "baseSeverity": "HIGH",
          "confidentialityImpact": "HIGH",
          "integrityImpact": "HIGH",
          "privilegesRequired": "LOW",
          "scope": "UNCHANGED",
          "userInteraction": "NONE",
          "vectorString": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
          "version": "3.1"
        },
        "exploitabilityScore": 2.8,
        "impactScore": 5.9,
        "source": "security-advisories@github.com",
        "type": "Secondary"
      }
    ]
  },
  "published": "2026-10-23T12:18:14.437",
  "references": [
    {
      "source": "security-advisories@github.com",
      "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m"
    }
  ],
  "sourceIdentifier": "security-advisories@github.com",
  "vulnStatus": "Deferred",
  "weaknesses": [
    {
      "description": [
        {
          "lang": "en",
          "value": "CWE-345"
        },
        {
          "lang": "en",
          "value": "CWE-863"
        }
      ],
      "source": "security-advisories@github.com",
      "type": "Primary"
    }
  ]
}

GHSA-8579-RGG5-PH2M

Vulnerability from github – Published: 2026-06-18 13:52 – Updated: 2026-07-20 21:23

Summary

PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Details

DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Summary

praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes.

The decision is not bound to:

a Discord reply to the approval message;
a Discord thread created for that request;
a Discord interaction/button callback for that request;
an explicit approver user allowlist; or
an approval nonce visible only to intended approvers.

As a result, any user who can post in the configured approval channel can approve a pending high-risk tool call by sending yes after the approval message appears. The same local PoV also shows that the Slack and Telegram messaging approval backends have no explicit approver allowlist parameter, but the primary report-grade issue is the Discord backend's unthreaded channel cross-talk: the approving message does not need to be a reply or otherwise request-bound.

Affected Product

Repository: MervinPraison/PraisonAI
Ecosystem: pip
Package: praisonai
Component: Python messaging approval backends
Primary affected file: src/praisonai/praisonai/bots/_discord_approval.py
Related sibling files:
src/praisonai/praisonai/bots/_slack_approval.py
src/praisonai/praisonai/bots/_telegram_approval.py
Latest PyPI version validated: 4.6.58
Current origin/main validated: 1ad58ca02975ff1398efeda694ea2ab78f20cf3e
Current origin/main tag validated: v4.6.58

Suggested affected range:

pip:praisonai >= 4.5.2, <= 4.6.58

Representative local sweep:

4.5.0    Discord approval backend not present
4.5.2    vulnerable
4.5.128  vulnerable
4.6.9    vulnerable
4.6.10   vulnerable
4.6.56   vulnerable
4.6.57   vulnerable
4.6.58   vulnerable

Root Cause

DiscordApproval.request_approval() posts an approval message to the configured channel and records the returned message id.

_poll_for_response() then polls channel history with:

f"/channels/{channel_id}/messages?after={message_id}&limit=10"

For each later non-bot message, it reads content, classifies the text, and returns ApprovalDecision(approved=True) when the text is approve/yes. There is no check that the message is a Discord reply to the approval message, belongs to a request-specific thread, came from an intended approver, or contains a request-specific approval token.

Important source evidence from origin/main:

_discord_approval.py lines 57-73: constructor accepts token, channel_id, timeout, and poll_interval; no approver allowlist.
_discord_approval.py line 239: polls all messages after the approval message in the configured channel.
_discord_approval.py lines 252-265: skips bot messages, classifies the remaining text, and approves when the keyword is approve.

The Slack and Telegram backends are better request-scoped than Discord:

Slack uses conversations.replies for the approval message thread.
Telegram checks the callback/reply message id.

However, both still lack an explicit approver identity parameter. They are included in the PoV and suggested fix because the authorization model should be consistent across all messaging approval backends.

Local PoV

Run against the latest local checkout:

python3 poc/pov_prai_cand_029_messaging_approval_channel_member_bypass.py \
  --repo ../../artifacts/repos/praisonai-v4.6.58 \
  --json

The PoV is local-only. It mocks Slack, Telegram, and Discord API helpers in-process and does not contact those services.

For Discord, the mock sequence is:

DiscordApproval posts a critical execute_command approval request to D_APPROVAL_CHANNEL.
The mocked channel-history endpoint returns a later ordinary non-bot channel message from D_INTRUDER with content yes.
DiscordApproval returns approved=True and approver="D_INTRUDER".

Observed output from evidence/pov-v4.6.58.json:

{
  "approval_from_unconfigured_channel_participant": {
    "discord": true,
    "slack": true,
    "telegram": true
  },
  "backends": [
    {
      "backend": "discord",
      "configured_channel_id": "D_APPROVAL_CHANNEL",
      "decision_approved": true,
      "decision_approver": "D_INTRUDER",
      "decision_reason": "Approved via Discord by intruder",
      "intruder_user": "D_INTRUDER"
    }
  ],
  "no_explicit_approver_allowlist_parameters": true,
  "vulnerable": true
}

The command in the approval request is a harmless local sentinel: touch /tmp/prai-cand-029. The PoV stops at the approval decision; it does not execute the tool.

Why This Is Not Intended Behavior

This report does not claim that a deliberately private approval channel with only trusted approvers is unsafe by itself. The narrower issue is that the Discord backend treats an unrelated later channel message as the approval decision for a specific dangerous tool request.

PraisonAI's approval documentation describes approval as a safety control that pauses before dangerous tools and asks a human or channel to allow or deny the specific request. A random later yes in the channel is not evidence that an intended approver reviewed that request.

The existing Slack and Telegram implementations already show request-binding patterns that Discord lacks:

Slack scopes to replies for the approval message timestamp.
Telegram checks the callback/reply message_id.

The Discord backend should provide at least the same request binding and should also support explicit approver identity checks for deployments where channel membership is broader than approval authority.

Impact

If an application uses DiscordApproval for dangerous tools such as shell commands, file writes, deletes, deployments, or other privileged operations, a low-privileged Discord user with write access to the configured approval channel can approve pending dangerous tool executions.

This can lead to code execution, file modification, deployment changes, or data access with the privileges of the PraisonAI process, depending on which tools the agent exposes behind approval.

The attacker does not need the LLM API key, shell access, repository access, or PraisonAI process access. They only need to be able to post an approval-looking message in the configured approval channel after the approval prompt appears.

Severity

Suggested severity: High.

Rationale:

AV: the attacker interacts through a networked Discord channel.
AC: sending yes after an approval prompt is enough.
PR: the attacker needs permission to post in the configured approval channel, but no approver-specific permission.
UI: no separate victim interaction is needed after the prompt exists.
S: the vulnerable approval backend and approved tool run in the PraisonAI application's security scope.
C/I/A: approved dangerous tools can disclose, modify, or destroy data depending on the configured agent tools.

Remediation

Recommended fixes:

For Discord, require approvals to be tied to the request, not merely any later channel message. Use Discord interactions/buttons with opaque server-side request ids, or require a Discord reply whose message_reference.message_id matches the approval message.
Add explicit approver identity configuration to all messaging approval backends, for example approver_user_ids or allowed_approvers.
Reject approvals from users outside the configured approver set, even if the message appears in the configured channel.
Include a per-request nonce or opaque approval id in callbacks and verify it server side before returning ApprovalDecision(approved=True).
Add regression tests for Discord where:
an unrelated later yes in the channel is ignored;
a reply from a non-approver is ignored;
a request-bound reply/callback from an allowed approver succeeds.

Severity

8.8 (High)


                  
                    CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

Show details on source website

JSON

To clipboard

{
  "affected": [
    {
      "database_specific": {
        "last_known_affected_version_range": "\u003c= 4.6.58"
      },
      "package": {
        "ecosystem": "PyPI",
        "name": "praisonai"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "4.5.2"
            },
            {
              "fixed": "4.6.59"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ]
    }
  ],
  "aliases": [
    "CVE-2026-56832"
  ],
  "database_specific": {
    "cwe_ids": [
      "CWE-345",
      "CWE-863"
    ],
    "github_reviewed": true,
    "github_reviewed_at": "2026-06-18T13:52:27Z",
    "nvd_published_at": null,
    "severity": "HIGH"
  },
  "details": "# DiscordApproval accepts unrelated channel messages as dangerous-tool approvals\n\n## Summary\n\n`praisonai.bots.DiscordApproval` approves a pending dangerous tool call when it\nsees any later non-bot message in the configured Discord channel whose text is\nclassified as approval, such as `yes`.\n\nThe decision is not bound to:\n\n- a Discord reply to the approval message;\n- a Discord thread created for that request;\n- a Discord interaction/button callback for that request;\n- an explicit approver user allowlist; or\n- an approval nonce visible only to intended approvers.\n\nAs a result, any user who can post in the configured approval channel can\napprove a pending high-risk tool call by sending `yes` after the approval\nmessage appears. The same local PoV also shows that the Slack and Telegram\nmessaging approval backends have no explicit approver allowlist parameter, but\nthe primary report-grade issue is the Discord backend\u0027s unthreaded channel\ncross-talk: the approving message does not need to be a reply or otherwise\nrequest-bound.\n\n## Affected Product\n\n- Repository: `MervinPraison/PraisonAI`\n- Ecosystem: `pip`\n- Package: `praisonai`\n- Component: Python messaging approval backends\n- Primary affected file: `src/praisonai/praisonai/bots/_discord_approval.py`\n- Related sibling files:\n  - `src/praisonai/praisonai/bots/_slack_approval.py`\n  - `src/praisonai/praisonai/bots/_telegram_approval.py`\n- Latest PyPI version validated: `4.6.58`\n- Current `origin/main` validated:\n  `1ad58ca02975ff1398efeda694ea2ab78f20cf3e`\n- Current `origin/main` tag validated: `v4.6.58`\n\nSuggested affected range:\n\n```text\npip:praisonai \u003e= 4.5.2, \u003c= 4.6.58\n```\n\nRepresentative local sweep:\n\n```text\n4.5.0    Discord approval backend not present\n4.5.2    vulnerable\n4.5.128  vulnerable\n4.6.9    vulnerable\n4.6.10   vulnerable\n4.6.56   vulnerable\n4.6.57   vulnerable\n4.6.58   vulnerable\n```\n\n## Root Cause\n\n`DiscordApproval.request_approval()` posts an approval message to the configured\nchannel and records the returned message id.\n\n`_poll_for_response()` then polls channel history with:\n\n```python\nf\"/channels/{channel_id}/messages?after={message_id}\u0026limit=10\"\n```\n\nFor each later non-bot message, it reads `content`, classifies the text, and\nreturns `ApprovalDecision(approved=True)` when the text is `approve`/`yes`.\nThere is no check that the message is a Discord reply to the approval message,\nbelongs to a request-specific thread, came from an intended approver, or\ncontains a request-specific approval token.\n\nImportant source evidence from `origin/main`:\n\n- `_discord_approval.py` lines 57-73: constructor accepts `token`,\n  `channel_id`, `timeout`, and `poll_interval`; no approver allowlist.\n- `_discord_approval.py` line 239: polls all messages after the approval\n  message in the configured channel.\n- `_discord_approval.py` lines 252-265: skips bot messages, classifies the\n  remaining text, and approves when the keyword is `approve`.\n\nThe Slack and Telegram backends are better request-scoped than Discord:\n\n- Slack uses `conversations.replies` for the approval message thread.\n- Telegram checks the callback/reply message id.\n\nHowever, both still lack an explicit approver identity parameter. They are\nincluded in the PoV and suggested fix because the authorization model should be\nconsistent across all messaging approval backends.\n\n## Local PoV\n\nRun against the latest local checkout:\n\n```bash\npython3 poc/pov_prai_cand_029_messaging_approval_channel_member_bypass.py \\\n  --repo ../../artifacts/repos/praisonai-v4.6.58 \\\n  --json\n```\n\nThe PoV is local-only. It mocks Slack, Telegram, and Discord API helpers\nin-process and does not contact those services.\n\nFor Discord, the mock sequence is:\n\n1. `DiscordApproval` posts a critical `execute_command` approval request to\n   `D_APPROVAL_CHANNEL`.\n2. The mocked channel-history endpoint returns a later ordinary non-bot\n   channel message from `D_INTRUDER` with content `yes`.\n3. `DiscordApproval` returns `approved=True` and `approver=\"D_INTRUDER\"`.\n\nObserved output from `evidence/pov-v4.6.58.json`:\n\n```json\n{\n  \"approval_from_unconfigured_channel_participant\": {\n    \"discord\": true,\n    \"slack\": true,\n    \"telegram\": true\n  },\n  \"backends\": [\n    {\n      \"backend\": \"discord\",\n      \"configured_channel_id\": \"D_APPROVAL_CHANNEL\",\n      \"decision_approved\": true,\n      \"decision_approver\": \"D_INTRUDER\",\n      \"decision_reason\": \"Approved via Discord by intruder\",\n      \"intruder_user\": \"D_INTRUDER\"\n    }\n  ],\n  \"no_explicit_approver_allowlist_parameters\": true,\n  \"vulnerable\": true\n}\n```\n\nThe command in the approval request is a harmless local sentinel:\n`touch /tmp/prai-cand-029`. The PoV stops at the approval decision; it does not\nexecute the tool.\n\n## Why This Is Not Intended Behavior\n\nThis report does not claim that a deliberately private approval channel with\nonly trusted approvers is unsafe by itself. The narrower issue is that the\nDiscord backend treats an unrelated later channel message as the approval\ndecision for a specific dangerous tool request.\n\nPraisonAI\u0027s approval documentation describes approval as a safety control that\npauses before dangerous tools and asks a human or channel to allow or deny the\nspecific request. A random later `yes` in the channel is not evidence that an\nintended approver reviewed that request.\n\nThe existing Slack and Telegram implementations already show request-binding\npatterns that Discord lacks:\n\n- Slack scopes to replies for the approval message timestamp.\n- Telegram checks the callback/reply `message_id`.\n\nThe Discord backend should provide at least the same request binding and should\nalso support explicit approver identity checks for deployments where channel\nmembership is broader than approval authority.\n\n## Impact\n\nIf an application uses `DiscordApproval` for dangerous tools such as shell\ncommands, file writes, deletes, deployments, or other privileged operations, a\nlow-privileged Discord user with write access to the configured approval channel\ncan approve pending dangerous tool executions.\n\nThis can lead to code execution, file modification, deployment changes, or data\naccess with the privileges of the PraisonAI process, depending on which tools\nthe agent exposes behind approval.\n\nThe attacker does not need the LLM API key, shell access, repository access, or\nPraisonAI process access. They only need to be able to post an approval-looking\nmessage in the configured approval channel after the approval prompt appears.\n\n## Severity\n\nSuggested severity: High.\n\nRationale:\n\n- `AV`: the attacker interacts through a networked Discord channel.\n- `AC`: sending `yes` after an approval prompt is enough.\n- `PR`: the attacker needs permission to post in the configured approval\n  channel, but no approver-specific permission.\n- `UI`: no separate victim interaction is needed after the prompt exists.\n- `S`: the vulnerable approval backend and approved tool run in the PraisonAI\n  application\u0027s security scope.\n- `C/I/A`: approved dangerous tools can disclose, modify, or destroy data\n  depending on the configured agent tools.\n\n## Remediation\n\nRecommended fixes:\n\n1. For Discord, require approvals to be tied to the request, not merely any\n   later channel message. Use Discord interactions/buttons with opaque\n   server-side request ids, or require a Discord reply whose\n   `message_reference.message_id` matches the approval message.\n2. Add explicit approver identity configuration to all messaging approval\n   backends, for example `approver_user_ids` or `allowed_approvers`.\n3. Reject approvals from users outside the configured approver set, even if the\n   message appears in the configured channel.\n4. Include a per-request nonce or opaque approval id in callbacks and verify it\n   server side before returning `ApprovalDecision(approved=True)`.\n5. Add regression tests for Discord where:\n   - an unrelated later `yes` in the channel is ignored;\n   - a reply from a non-approver is ignored;\n   - a request-bound reply/callback from an allowed approver succeeds.",
  "id": "GHSA-8579-rgg5-ph2m",
  "modified": "2026-07-20T21:23:11Z",
  "published": "2026-06-18T13:52:27Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/MervinPraison/PraisonAI"
    }
  ],
  "schema_version": "1.4.0",
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
      "type": "CVSS_V3"
    }
  ],
  "summary": "PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals"
}

PYSEC-2026-3506

Vulnerability from pysec - Published: 2026-07-23 11:41 - Updated: 2026-07-23 14:32

Details

DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Summary

praisonai.bots.DiscordApproval approves a pending dangerous tool call when it sees any later non-bot message in the configured Discord channel whose text is classified as approval, such as yes.

The decision is not bound to:

a Discord reply to the approval message;
a Discord thread created for that request;
a Discord interaction/button callback for that request;
an explicit approver user allowlist; or
an approval nonce visible only to intended approvers.

Affected Product

Repository: MervinPraison/PraisonAI
Ecosystem: pip
Package: praisonai
Component: Python messaging approval backends
Primary affected file: src/praisonai/praisonai/bots/_discord_approval.py
Related sibling files:
src/praisonai/praisonai/bots/_slack_approval.py
src/praisonai/praisonai/bots/_telegram_approval.py
Latest PyPI version validated: 4.6.58
Current origin/main validated: 1ad58ca02975ff1398efeda694ea2ab78f20cf3e
Current origin/main tag validated: v4.6.58

Suggested affected range:

pip:praisonai >= 4.5.2, <= 4.6.58

Representative local sweep:

4.5.0    Discord approval backend not present
4.5.2    vulnerable
4.5.128  vulnerable
4.6.9    vulnerable
4.6.10   vulnerable
4.6.56   vulnerable
4.6.57   vulnerable
4.6.58   vulnerable

Root Cause

DiscordApproval.request_approval() posts an approval message to the configured channel and records the returned message id.

_poll_for_response() then polls channel history with:

f"/channels/{channel_id}/messages?after={message_id}&limit=10"

Important source evidence from origin/main:

_discord_approval.py lines 57-73: constructor accepts token, channel_id, timeout, and poll_interval; no approver allowlist.
_discord_approval.py line 239: polls all messages after the approval message in the configured channel.
_discord_approval.py lines 252-265: skips bot messages, classifies the remaining text, and approves when the keyword is approve.

The Slack and Telegram backends are better request-scoped than Discord:

Slack uses conversations.replies for the approval message thread.
Telegram checks the callback/reply message id.

Local PoV

Run against the latest local checkout:

python3 poc/pov_prai_cand_029_messaging_approval_channel_member_bypass.py \
  --repo ../../artifacts/repos/praisonai-v4.6.58 \
  --json

The PoV is local-only. It mocks Slack, Telegram, and Discord API helpers in-process and does not contact those services.

For Discord, the mock sequence is:

DiscordApproval posts a critical execute_command approval request to D_APPROVAL_CHANNEL.
The mocked channel-history endpoint returns a later ordinary non-bot channel message from D_INTRUDER with content yes.
DiscordApproval returns approved=True and approver="D_INTRUDER".

Observed output from evidence/pov-v4.6.58.json:

{
  "approval_from_unconfigured_channel_participant": {
    "discord": true,
    "slack": true,
    "telegram": true
  },
  "backends": [
    {
      "backend": "discord",
      "configured_channel_id": "D_APPROVAL_CHANNEL",
      "decision_approved": true,
      "decision_approver": "D_INTRUDER",
      "decision_reason": "Approved via Discord by intruder",
      "intruder_user": "D_INTRUDER"
    }
  ],
  "no_explicit_approver_allowlist_parameters": true,
  "vulnerable": true
}

The command in the approval request is a harmless local sentinel: touch /tmp/prai-cand-029. The PoV stops at the approval decision; it does not execute the tool.

Why This Is Not Intended Behavior

The existing Slack and Telegram implementations already show request-binding patterns that Discord lacks:

Slack scopes to replies for the approval message timestamp.
Telegram checks the callback/reply message_id.

Impact

This can lead to code execution, file modification, deployment changes, or data access with the privileges of the PraisonAI process, depending on which tools the agent exposes behind approval.

Severity

Suggested severity: High.

Rationale:

AV: the attacker interacts through a networked Discord channel.
AC: sending yes after an approval prompt is enough.
PR: the attacker needs permission to post in the configured approval channel, but no approver-specific permission.
UI: no separate victim interaction is needed after the prompt exists.
S: the vulnerable approval backend and approved tool run in the PraisonAI application's security scope.
C/I/A: approved dangerous tools can disclose, modify, or destroy data depending on the configured agent tools.

Remediation

Recommended fixes:

For Discord, require approvals to be tied to the request, not merely any later channel message. Use Discord interactions/buttons with opaque server-side request ids, or require a Discord reply whose message_reference.message_id matches the approval message.
Add explicit approver identity configuration to all messaging approval backends, for example approver_user_ids or allowed_approvers.
Reject approvals from users outside the configured approver set, even if the message appears in the configured channel.
Include a per-request nonce or opaque approval id in callbacks and verify it server side before returning ApprovalDecision(approved=True).
Add regression tests for Discord where:
an unrelated later yes in the channel is ignored;
a reply from a non-approver is ignored;
a request-bound reply/callback from an allowed approver succeeds.

Severity

8.8 (High)


                  
                    CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H

Impacted products

Name	purl
praisonai	pkg:pypi/praisonai

Aliases

JSON

To clipboard

{
  "affected": [
    {
      "package": {
        "ecosystem": "PyPI",
        "name": "praisonai",
        "purl": "pkg:pypi/praisonai"
      },
      "ranges": [
        {
          "events": [
            {
              "introduced": "4.5.2"
            },
            {
              "fixed": "4.6.59"
            }
          ],
          "type": "ECOSYSTEM"
        }
      ],
      "versions": [
        "4.5.10",
        "4.5.100",
        "4.5.101",
        "4.5.102",
        "4.5.103",
        "4.5.104",
        "4.5.105",
        "4.5.106",
        "4.5.107",
        "4.5.108",
        "4.5.109",
        "4.5.11",
        "4.5.110",
        "4.5.111",
        "4.5.112",
        "4.5.113",
        "4.5.114",
        "4.5.115",
        "4.5.117",
        "4.5.118",
        "4.5.119",
        "4.5.12",
        "4.5.120",
        "4.5.121",
        "4.5.122",
        "4.5.123",
        "4.5.124",
        "4.5.125",
        "4.5.126",
        "4.5.127",
        "4.5.128",
        "4.5.129",
        "4.5.13",
        "4.5.130",
        "4.5.131",
        "4.5.132",
        "4.5.133",
        "4.5.134",
        "4.5.135",
        "4.5.136",
        "4.5.137",
        "4.5.139",
        "4.5.14",
        "4.5.140",
        "4.5.143",
        "4.5.144",
        "4.5.145",
        "4.5.149",
        "4.5.15",
        "4.5.16",
        "4.5.18",
        "4.5.19",
        "4.5.2",
        "4.5.20",
        "4.5.21",
        "4.5.22",
        "4.5.23",
        "4.5.24",
        "4.5.25",
        "4.5.26",
        "4.5.27",
        "4.5.28",
        "4.5.29",
        "4.5.3",
        "4.5.30",
        "4.5.31",
        "4.5.32",
        "4.5.33",
        "4.5.34",
        "4.5.35",
        "4.5.36",
        "4.5.37",
        "4.5.38",
        "4.5.39",
        "4.5.40",
        "4.5.41",
        "4.5.42",
        "4.5.43",
        "4.5.44",
        "4.5.45",
        "4.5.46",
        "4.5.48",
        "4.5.49",
        "4.5.5",
        "4.5.51",
        "4.5.52",
        "4.5.54",
        "4.5.55",
        "4.5.56",
        "4.5.57",
        "4.5.58",
        "4.5.59",
        "4.5.6",
        "4.5.60",
        "4.5.62",
        "4.5.63",
        "4.5.64",
        "4.5.65",
        "4.5.67",
        "4.5.68",
        "4.5.69",
        "4.5.7",
        "4.5.70",
        "4.5.71",
        "4.5.72",
        "4.5.73",
        "4.5.74",
        "4.5.76",
        "4.5.77",
        "4.5.78",
        "4.5.79",
        "4.5.8",
        "4.5.80",
        "4.5.81",
        "4.5.82",
        "4.5.83",
        "4.5.85",
        "4.5.87",
        "4.5.88",
        "4.5.89",
        "4.5.9",
        "4.5.90",
        "4.5.93",
        "4.5.94",
        "4.5.95",
        "4.5.96",
        "4.5.97",
        "4.5.98",
        "4.6.10",
        "4.6.11",
        "4.6.12",
        "4.6.13",
        "4.6.14",
        "4.6.15",
        "4.6.16",
        "4.6.18",
        "4.6.19",
        "4.6.20",
        "4.6.21",
        "4.6.22",
        "4.6.23",
        "4.6.24",
        "4.6.25",
        "4.6.26",
        "4.6.27",
        "4.6.28",
        "4.6.29",
        "4.6.30",
        "4.6.31",
        "4.6.32",
        "4.6.33",
        "4.6.34",
        "4.6.35",
        "4.6.36",
        "4.6.37",
        "4.6.38",
        "4.6.39",
        "4.6.40",
        "4.6.41",
        "4.6.42",
        "4.6.43",
        "4.6.44",
        "4.6.45",
        "4.6.46",
        "4.6.47",
        "4.6.48",
        "4.6.50",
        "4.6.51",
        "4.6.52",
        "4.6.53",
        "4.6.54",
        "4.6.55",
        "4.6.56",
        "4.6.57",
        "4.6.58",
        "4.6.9"
      ]
    }
  ],
  "aliases": [
    "CVE-2026-56832",
    "GHSA-8579-rgg5-ph2m"
  ],
  "details": "# DiscordApproval accepts unrelated channel messages as dangerous-tool approvals\n\n## Summary\n\n`praisonai.bots.DiscordApproval` approves a pending dangerous tool call when it\nsees any later non-bot message in the configured Discord channel whose text is\nclassified as approval, such as `yes`.\n\nThe decision is not bound to:\n\n- a Discord reply to the approval message;\n- a Discord thread created for that request;\n- a Discord interaction/button callback for that request;\n- an explicit approver user allowlist; or\n- an approval nonce visible only to intended approvers.\n\nAs a result, any user who can post in the configured approval channel can\napprove a pending high-risk tool call by sending `yes` after the approval\nmessage appears. The same local PoV also shows that the Slack and Telegram\nmessaging approval backends have no explicit approver allowlist parameter, but\nthe primary report-grade issue is the Discord backend\u0027s unthreaded channel\ncross-talk: the approving message does not need to be a reply or otherwise\nrequest-bound.\n\n## Affected Product\n\n- Repository: `MervinPraison/PraisonAI`\n- Ecosystem: `pip`\n- Package: `praisonai`\n- Component: Python messaging approval backends\n- Primary affected file: `src/praisonai/praisonai/bots/_discord_approval.py`\n- Related sibling files:\n  - `src/praisonai/praisonai/bots/_slack_approval.py`\n  - `src/praisonai/praisonai/bots/_telegram_approval.py`\n- Latest PyPI version validated: `4.6.58`\n- Current `origin/main` validated:\n  `1ad58ca02975ff1398efeda694ea2ab78f20cf3e`\n- Current `origin/main` tag validated: `v4.6.58`\n\nSuggested affected range:\n\n```text\npip:praisonai \u003e= 4.5.2, \u003c= 4.6.58\n```\n\nRepresentative local sweep:\n\n```text\n4.5.0    Discord approval backend not present\n4.5.2    vulnerable\n4.5.128  vulnerable\n4.6.9    vulnerable\n4.6.10   vulnerable\n4.6.56   vulnerable\n4.6.57   vulnerable\n4.6.58   vulnerable\n```\n\n## Root Cause\n\n`DiscordApproval.request_approval()` posts an approval message to the configured\nchannel and records the returned message id.\n\n`_poll_for_response()` then polls channel history with:\n\n```python\nf\"/channels/{channel_id}/messages?after={message_id}\u0026limit=10\"\n```\n\nFor each later non-bot message, it reads `content`, classifies the text, and\nreturns `ApprovalDecision(approved=True)` when the text is `approve`/`yes`.\nThere is no check that the message is a Discord reply to the approval message,\nbelongs to a request-specific thread, came from an intended approver, or\ncontains a request-specific approval token.\n\nImportant source evidence from `origin/main`:\n\n- `_discord_approval.py` lines 57-73: constructor accepts `token`,\n  `channel_id`, `timeout`, and `poll_interval`; no approver allowlist.\n- `_discord_approval.py` line 239: polls all messages after the approval\n  message in the configured channel.\n- `_discord_approval.py` lines 252-265: skips bot messages, classifies the\n  remaining text, and approves when the keyword is `approve`.\n\nThe Slack and Telegram backends are better request-scoped than Discord:\n\n- Slack uses `conversations.replies` for the approval message thread.\n- Telegram checks the callback/reply message id.\n\nHowever, both still lack an explicit approver identity parameter. They are\nincluded in the PoV and suggested fix because the authorization model should be\nconsistent across all messaging approval backends.\n\n## Local PoV\n\nRun against the latest local checkout:\n\n```bash\npython3 poc/pov_prai_cand_029_messaging_approval_channel_member_bypass.py \\\n  --repo ../../artifacts/repos/praisonai-v4.6.58 \\\n  --json\n```\n\nThe PoV is local-only. It mocks Slack, Telegram, and Discord API helpers\nin-process and does not contact those services.\n\nFor Discord, the mock sequence is:\n\n1. `DiscordApproval` posts a critical `execute_command` approval request to\n   `D_APPROVAL_CHANNEL`.\n2. The mocked channel-history endpoint returns a later ordinary non-bot\n   channel message from `D_INTRUDER` with content `yes`.\n3. `DiscordApproval` returns `approved=True` and `approver=\"D_INTRUDER\"`.\n\nObserved output from `evidence/pov-v4.6.58.json`:\n\n```json\n{\n  \"approval_from_unconfigured_channel_participant\": {\n    \"discord\": true,\n    \"slack\": true,\n    \"telegram\": true\n  },\n  \"backends\": [\n    {\n      \"backend\": \"discord\",\n      \"configured_channel_id\": \"D_APPROVAL_CHANNEL\",\n      \"decision_approved\": true,\n      \"decision_approver\": \"D_INTRUDER\",\n      \"decision_reason\": \"Approved via Discord by intruder\",\n      \"intruder_user\": \"D_INTRUDER\"\n    }\n  ],\n  \"no_explicit_approver_allowlist_parameters\": true,\n  \"vulnerable\": true\n}\n```\n\nThe command in the approval request is a harmless local sentinel:\n`touch /tmp/prai-cand-029`. The PoV stops at the approval decision; it does not\nexecute the tool.\n\n## Why This Is Not Intended Behavior\n\nThis report does not claim that a deliberately private approval channel with\nonly trusted approvers is unsafe by itself. The narrower issue is that the\nDiscord backend treats an unrelated later channel message as the approval\ndecision for a specific dangerous tool request.\n\nPraisonAI\u0027s approval documentation describes approval as a safety control that\npauses before dangerous tools and asks a human or channel to allow or deny the\nspecific request. A random later `yes` in the channel is not evidence that an\nintended approver reviewed that request.\n\nThe existing Slack and Telegram implementations already show request-binding\npatterns that Discord lacks:\n\n- Slack scopes to replies for the approval message timestamp.\n- Telegram checks the callback/reply `message_id`.\n\nThe Discord backend should provide at least the same request binding and should\nalso support explicit approver identity checks for deployments where channel\nmembership is broader than approval authority.\n\n## Impact\n\nIf an application uses `DiscordApproval` for dangerous tools such as shell\ncommands, file writes, deletes, deployments, or other privileged operations, a\nlow-privileged Discord user with write access to the configured approval channel\ncan approve pending dangerous tool executions.\n\nThis can lead to code execution, file modification, deployment changes, or data\naccess with the privileges of the PraisonAI process, depending on which tools\nthe agent exposes behind approval.\n\nThe attacker does not need the LLM API key, shell access, repository access, or\nPraisonAI process access. They only need to be able to post an approval-looking\nmessage in the configured approval channel after the approval prompt appears.\n\n## Severity\n\nSuggested severity: High.\n\nRationale:\n\n- `AV`: the attacker interacts through a networked Discord channel.\n- `AC`: sending `yes` after an approval prompt is enough.\n- `PR`: the attacker needs permission to post in the configured approval\n  channel, but no approver-specific permission.\n- `UI`: no separate victim interaction is needed after the prompt exists.\n- `S`: the vulnerable approval backend and approved tool run in the PraisonAI\n  application\u0027s security scope.\n- `C/I/A`: approved dangerous tools can disclose, modify, or destroy data\n  depending on the configured agent tools.\n\n## Remediation\n\nRecommended fixes:\n\n1. For Discord, require approvals to be tied to the request, not merely any\n   later channel message. Use Discord interactions/buttons with opaque\n   server-side request ids, or require a Discord reply whose\n   `message_reference.message_id` matches the approval message.\n2. Add explicit approver identity configuration to all messaging approval\n   backends, for example `approver_user_ids` or `allowed_approvers`.\n3. Reject approvals from users outside the configured approver set, even if the\n   message appears in the configured channel.\n4. Include a per-request nonce or opaque approval id in callbacks and verify it\n   server side before returning `ApprovalDecision(approved=True)`.\n5. Add regression tests for Discord where:\n   - an unrelated later `yes` in the channel is ignored;\n   - a reply from a non-approver is ignored;\n   - a request-bound reply/callback from an allowed approver succeeds.",
  "id": "PYSEC-2026-3506",
  "modified": "2026-07-23T14:32:29.962648Z",
  "published": "2026-07-23T11:41:39.857428Z",
  "references": [
    {
      "type": "WEB",
      "url": "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-8579-rgg5-ph2m"
    },
    {
      "type": "PACKAGE",
      "url": "https://github.com/MervinPraison/PraisonAI"
    },
    {
      "type": "PACKAGE",
      "url": "https://pypi.org/project/praisonai"
    },
    {
      "type": "ADVISORY",
      "url": "https://github.com/advisories/GHSA-8579-rgg5-ph2m"
    },
    {
      "type": "ADVISORY",
      "url": "https://nvd.nist.gov/vuln/detail/CVE-2026-56832"
    }
  ],
  "severity": [
    {
      "score": "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:H",
      "type": "CVSS_V3"
    }
  ],
  "summary": "PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals"
}

Sightings

Author	Source	Type	Date	Other

Nomenclature

Seen: The vulnerability was mentioned, discussed, or observed by the user.
Confirmed: The vulnerability has been validated from an analyst's perspective.
Published Proof of Concept: A public proof of concept is available for this vulnerability.
Exploited: The vulnerability was observed as exploited by the user who reported the sighting.
Patched: The vulnerability was observed as successfully patched by the user who reported the sighting.
Not exploited: The vulnerability was not observed as exploited by the user who reported the sighting.
Not confirmed: The user expressed doubt about the validity of the vulnerability.
Not patched: The vulnerability was not observed as successfully patched by the user who reported the sighting.

Vulnerability timeline

Sightings and recorded dates, grouped by day · UTC

Order JSON

Publication & reservation Sightings KEV catalogs Other timestamps

Loading timeline…

The MITRE ATT&CK techniques below are AI-generated suggestions, inferred from the description of the vulnerability by the CIRCL/vulnerability-attack-technique-classification-roberta-base model, served locally by ML-Gateway. They have not been verified by an analyst and are provided for guidance only.
The approach is described in our paper Mapping CVEs to MITRE ATT&CK Techniques: A Curated Gold-Set Classifier and the Limits of LLM-Assisted Label Expansion.
Browse all ATT&CK techniques and the vulnerabilities related to each.

Related by attack behaviour

Vulnerabilities whose description is nearest to this one in the vector space of the CIRCL/vulnerability-attack-technique-biencoder model. This is a similarity search over the bi-encoder space (plain cosine), not a classification, and it has no measured accuracy.

Action not permitted

CVE-2026-56832 (GCVE-0-2026-56832)

FKIE_CVE-2026-56832

GHSA-8579-RGG5-PH2M

DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Summary

Affected Product

Root Cause

Local PoV

Why This Is Not Intended Behavior

Impact

Severity

Remediation

PYSEC-2026-3506

DiscordApproval accepts unrelated channel messages as dangerous-tool approvals

Summary

Affected Product

Root Cause

Local PoV

Why This Is Not Intended Behavior

Impact

Severity

Remediation

Tags

Sightings

Nomenclature

Vulnerability timeline

Related by attack behaviour