PR
Product
PraisonAI
147
CVEs
0
On KEV
10.0
Max CVSS
18
New 30D
Severity mix
Critical 43High 83Medium 21Low 0
Watch
Add PraisonAI to your stack to get only the CVEs that touch what you run.
| CVE | CVSS | Published | Summary · affected versions |
|---|---|---|---|
| CVE-2026-48168 | 10.0 | Aug 5 | PraisonAI: GitHub Actions Claude workflow command injection via unquoted PR branch name Affected: < 4.6.40 |
| CVE-2026-61447 | 10.0 | Jul 11 | PraisonAI before 1.6.78 Remote Code Execution via CodeAgent Affected: before 1.6.78 |
| CVE-2026-40114 | 10.0 | Apr 9 | PraisonAI has Server-Side Request Forgery via Unvalidated webhook_url in Jobs API Affected: < 4.5.128 |
| CVE-2026-39305 | 10.0 | Apr 7 | Arbitrary File Write / Path Traversal in Action Orchestrator Affected: < 4.5.113 |
| CVE-2026-34955 | 10.0 | Apr 3 | PraisonAI: Sandbox Escape via shell=True and Bypassable Blocklist in SubprocessSandbox Affected: < 4.5.97 |
| CVE-2026-34938 | 10.0 | Apr 3 | PraisonAI: Python Sandbox Escape via str Subclass startswith() Override in execute_code Affected: < 1.5.90 |
| CVE-2026-57138 | 9.9 | Sep 15 | PraisonAI codeMode sandbox escape via Function constructor Affected: >= 1.4.0, < 1.7.2 |
| CVE-2026-47392 | 9.9 | Jul 21 | PraisonAI vulnerable to sandbox escape via `print.__self__` builtins module leak in `execute_code` (subprocess mode) Affected: < 4.6.40 |
| CVE-2026-61445 | 9.9 | Jul 11 | PraisonAI before 4.6.78 Arbitrary File Write and Command Execution Affected: before 4.6.78 |
| CVE-2026-57116 New | 9.8 | Oct 23 | PraisonAI: AgentOS remains unauthenticated after GHSA-pm96 patched version and allows remote agent invocation Affected: >= 4.2.1, < 4.6.59 |
| CVE-2026-57118 | 9.8 | Oct 12 | PraisonAI AgentTeam.launch exposes unauthenticated remote agent listing and invocation endpoints Affected: < 4.6.59 |
| CVE-2026-57147 | 9.8 | Sep 15 | praisonai-platform: default JWT signing secret 'dev-secret-change-me' enables token forgery Affected: < 4.6.51 |
| CVE-2026-57148 | 9.8 | Sep 15 | praisonai-platform 0.1.4 still boots on the hardcoded JWT secret dev-secret-change-me (default-open production guard) Affected: < 4.6.51 |
| CVE-2026-57139 | 9.8 | Sep 15 | PraisonAI MCPServer exposes unauthenticated HTTP tools/call Affected: >= 1.5.0, < 1.7.2 |
| CVE-2026-57141 | 9.8 | Sep 15 | PraisonAI: Remote Code Execution via Sandbox Escape in `codeMode` Tool Affected: < 1.7.2 |
| CVE-2026-57131 | 9.8 | Sep 14 | praisonai: Jobs API exposes agent-execution endpoints with no authentication Affected: < 4.6.48 |
| CVE-2026-57124 | 9.8 | Sep 14 | PraisonAI UI MCP connect endpoint allows unauthenticated local command execution Affected: < 4.6.59 |
| CVE-2026-57127 | 9.8 | Sep 14 | praisonai: recipe serve auth middleware silently disables itself when no secret is set Affected: < 4.6.48 |
| CVE-2026-57125 | 9.8 | Sep 14 | PraisonAI: Unauthenticated RCE via Jobs API + Approval Bypass Affected: < 4.6.59 |
| CVE-2026-47396 | 9.8 | Jul 21 | PraisonAI call server exposes unauthenticated agent listing, invocation, and deletion when CALL_SERVER_TOKEN is unset Affected: < 4.6.40 |
| CVE-2026-47393 | 9.8 | Jul 21 | PraisonAI `deploy --type api` emits a Flask server with authentication disabled by default Affected: < 4.6.40 |
| CVE-2026-47391 | 9.8 | Jul 21 | PraisonAI's unauthenticated A2A official example can reach real LLM-driven `eval()` tool execution Affected: < 4.6.40 |
| CVE-2026-60090 | 9.8 | Jul 11 | PraisonAI before 4.6.78 SQL/CQL Injection via vector dimension Affected: before 4.6.78 |
| CVE-2026-44335 | 9.8 | May 8 | SSRF bypass in PraisonAI Affected: < 1.6.32 |
| CVE-2026-41497 | 9.8 | May 8 | Incomplete fix for CVE-2026-34935: Command Injection in MervinPraison/PraisonAI Affected: < 4.6.9 |
| CVE-2026-40288 | 9.8 | Apr 14 | PraisonAI: Critical RCE via `type: job` workflow YAML Affected: < 4.5.139 |
| CVE-2026-40315 | 9.8 | Apr 14 | PraisonAI: SQLiteConversationStore didn't validate table_prefix when constructing SQL queries Affected: < 4.5.133 |
| CVE-2026-39890 | 9.8 | Apr 8 | PraisonAI Affected by Remote Code Execution via YAML Deserialization in Agent Definition Loading Affected: < 4.5.115 |
| CVE-2026-34937 | 9.8 | Apr 3 | PraisonAI: Shell Injection in run_python() via Unescaped $() Substitution Affected: < 1.5.90 |
| CVE-2026-34934 | 9.8 | Apr 3 | PraisonAI: Second-Order SQL Injection in `get_all_user_threads` Affected: < 4.5.90 |
| CVE-2026-34935 | 9.8 | Apr 3 | PraisonAI: OS Command Injection in MCPHandler.parse_mcp_command() Affected: >= 4.5.15, < 4.5.69 |
| CVE-2026-44336 | 9.6 | May 8 | PraisonAI MCP `tools/call` path-traversal and RCE via Python `.pth` injection Affected: < 4.6.34 |
| CVE-2026-40154 | 9.6 | Apr 9 | PraisonAI Affected by Untrusted Remote Template Code Execution Affected: < 4.5.128 |
| CVE-2026-40088 | 9.6 | Apr 9 | Improper Neutralization of Special Elements used in an OS Command ('OS Command Injection') in praisonai Affected: < 4.5.121 |
| CVE-2026-57140 | 9.4 | Sep 15 | PraisonAI AgentOS exposes unauthenticated agent listing and invocation Affected: >= 1.6.0, < 1.7.2 |
| CVE-2026-62176 | 9.1 | Oct 7 | PraisonAI: Code Injection via f-string Interpolation in Deploy API Server Generation Affected: < 4.6.78 |
| CVE-2026-57145 | 9.1 | Sep 14 | PraisonAI: Arbitrary File Read/Write via `multiedit` Tool Without Path Validation Affected: < 4.6.62 |
| CVE-2026-55536 | 9.1 | Aug 25 | Browser Server WebSocket origin validation bypass via unanchored regex (patch bypass of CVE-2026-40289 / GHSA-8x8f-54wf-vv92) Affected: < 4.6.58 |
| CVE-2026-61444 | 9.1 | Jul 10 | PraisonAI before 4.6.78 Code Injection via f-string Affected: before 4.6.78 |
| CVE-2026-40313 | 9.1 | Apr 14 | PraisonAI: ArtiPACKED Vulnerability via GitHub Actions Credential Persistence Affected: < 4.5.140 |
| CVE-2026-40289 | 9.1 | Apr 14 | PraisonAI Browser Server allows unauthenticated WebSocket clients to hijack connected extension sessions Affected: < 4.5.139 |
| CVE-2026-34953 | 9.1 | Apr 3 | PraisonAI: Authentication Bypass in OAuthManager.validate_token() Affected: < 4.5.97 |
| CVE-2026-34952 | 9.1 | Apr 3 | PraisonAI: Missing Authentication in WebSocket Gateway Affected: < 4.5.97 |
| CVE-2026-57117 New | 8.8 | Oct 23 | PraisonAI: Compute-bridged file tools allow shell command injection Affected: >= 4.6.10, < 4.6.59 |
| CVE-2026-57144 New | 8.8 | Oct 23 | PraisonAI SandlockSandbox falls back to unrestricted subprocess execution when Landlock is unavailable Affected: >= 4.5.110, < 4.6.61 |
| CVE-2026-56840 New | 8.8 | Oct 23 | PraisonAI: HTTPApproval dashboard renders tool arguments as raw HTML, allowing approval-page XSS to approve dangerous tools Affected: >= 4.5.2, < 4.6.59 |
| CVE-2026-56832 New | 8.8 | Oct 23 | PraisonAI DiscordApproval accepts unrelated channel messages as dangerous-tool approvals Affected: >= 4.5.2, < 4.6.59 |
| CVE-2026-57133 | 8.8 | Sep 15 | PraisonAI utility shell safe-command wrapper allowlist bypass via shell chaining Affected: >= 1.5.1, < 1.7.2 |
| CVE-2026-57136 | 8.8 | Sep 15 | PraisonAI SandboxExecutor allowedCommands bypass via shell chaining Affected: >= 1.2.3, < 1.7.2 |
| CVE-2026-57137 | 8.8 | Sep 15 | PraisonAI AgentLoop onToolCall approval runs after tool execution Affected: >= 1.4.0, < 1.7.2 |
