These pages are published after PatchSiren validates generated defensive summaries against stored public CVE and source evidence.
CVE-2026-61442 is a high-severity vulnerability in PraisonAI Platform before version 0.1.9. The vulnerability allows a workspace member to bypass owner/admin authorization on PATCH routes for projects, issues, and agents, potentially leading to unauthorized modifications and deletions. This could result in the reassignment of lead_id to a workspace member's user id and subsequent deletion of owner-created [truncated]
CVE-2026-61429 is a high-severity server-side request forgery vulnerability in PraisonAI versions before 1.6.78. The vulnerability is located in the Crawl4AI/Chromium backend and allows attackers to bypass SSRF validation by exploiting DNS rebinding and HTTP redirects. This vulnerability enables attackers to craft URLs that resolve to internal services after the initial validation check, allowing the head [truncated]
CVE-2026-61428 is a medium-severity vulnerability in PraisonAI AgentMail versions before 4.6.78. The vulnerability allows unauthenticated attackers to inject messages with spoofed sender addresses by exploiting the lack of signature verification in webhook mode. This can lead to arbitrary content being injected into the agent and replies being triggered to attacker-controlled addresses, bypassing sender a [truncated]
CVE-2026-56078 is a high-severity path traversal vulnerability in PraisonAI's MultiAgentMonitor. Attackers can exploit this vulnerability to read, write, or overwrite arbitrary files, potentially leading to sensitive disclosure, denial of service, or code execution. The vulnerability has a CVSS score of 8.7 and is considered HIGH severity. PraisonAI versions before 1.5.115 are affected. Users should updat [truncated]
CVE-2026-56077 is a high-severity information disclosure vulnerability in the MultiAgentLedger component of PraisonAI versions before 1.5.115. This vulnerability allows attackers to access sensitive data by registering agents with duplicate IDs, exploiting the lack of agent ID uniqueness enforcement. Consequently, attackers can share ledger instances and expose system prompts and conversation history betw [truncated]
CVE-2026-56076 is a high-severity vulnerability in PraisonAI, a tool with an unknown vendor, that allows remote attackers to execute arbitrary agent actions. This is possible because the AGUI endpoint lacks authentication and has hardcoded wildcard CORS headers. Specifically, the POST /agui endpoint does not require authentication and includes 'Access-Control-Allow-Origin: *' headers. When combined with S [truncated]
CVE-2026-56075 is an arbitrary shell command execution vulnerability in PraisonAI before version 4.5.128. The vulnerability arises from UI modules hardcoding approval_mode to auto, which overrides administrator configuration from the PRAISON_APPROVAL_MODE environment variable. Authenticated attackers can instruct the LLM agent to execute arbitrary shell commands via subprocess.run with shell=True, bypassi [truncated]
CVE-2026-56074 is a medium-severity vulnerability in PraisonAI that allows attackers to bypass tool approval prompts by exploiting coarse-grained caching of tool approval decisions. The vulnerability exists because PraisonAI caches tool approval decisions by tool name only, not by invocation arguments. This means that an attacker can obtain initial approval for a benign command and then silently exfiltrat [truncated]