CVE-2026-57130: PraisonAI IMAP Command Injection Vulnerability

{
  "schema_version" : "1.4.0",
  "id" : "GHSA-c969-5x3p-vq3v",
  "modified" : "2026-07-20T21:26:50Z",
  "published" : "2026-06-18T14:25:03Z",
  "aliases" : [ "CVE-2026-57130" ],
  "summary" : "PraisonAI: IMAP Command Injection via Unsanitized Email Search Parameters",
  "details" : "## Summary\n\nThe email search tool in `src/praisonai-agents/praisonaiagents/tools/email_tools.py` constructs IMAP SEARCH commands by interpolating LLM-controlled parameters (from_addr, subject, query) directly into IMAP protocol strings using f-string formatting with double-quote delimiters. An attacker who can influence the arguments to the `search_emails` or `reply_email` tool (via crafted agent prompts) can inject arbitrary IMAP commands, potentially exfiltrating email data from other folders, deleting emails, or performing other unauthorized IMAP operations.\n## Details\n\n**Vulnerable code (lines 493–502):**\n```python\ncriteria = []\nif from_addr:\n    criteria.append(f'FROM \"{from_addr}\"')\nif subject:\n    criteria.append(f'SUBJECT \"{subject}\"')\nif query:\n    criteria.append(f'TEXT \"{query}\"')\nif not criteria:\n    criteria.append(\"ALL\")\nsearch_str = \" \".join(criteria)\nstatus, data = mail.search(None, search_str)\n```\n\nThe `from_addr`, `subject`, and `query` parameters originate from LLM tool call arguments (the `search_emails` public function at line 665). These values flow through without any sanitization or escaping. The double-quote (`\"`) characters in these parameters allow breaking out of the IMAP SEARCH quoted string context.\n\n**Additional injection points:**\n- Line 416: `mail.search(None, f'HEADER Message-ID \"{search_id}\"')`\n- Line 447: Same pattern in `_smtp_reply_email`\n- Line 542: Same pattern in `_smtp_archive_email`\n\nThe `search_id` / `message_id` parameter in these functions is also LLM-controlled via the `reply_email` and `archive_email` public tool functions.\n\n**Reachability:** The `search_emails`, `reply_email`, and `archive_email` functions are exposed as agent tools. They are reachable when an agent is configured with email tools (EMAIL_ADDRESS + EMAIL_PASSWORD environment variables set). This is a documented deployment scenario for email-capable agents.\n\n## PoC\n\n**Setup:** Requires an IMAP server (not run here — this is a static proof). The vulnerability is demonstrated by tracing the data flow.\n\n**Positive trigger — IMAP injection via `search_emails`:**\nAn LLM agent processing a crafted prompt calls:\n```python\nsearch_emails(from_addr='user@example.com\" LOGOUT')\n```\nThis produces the IMAP command:\n```\nSEARCH FROM \"user@example.com\" LOGOUT\"\n```\nThe `LOGOUT` command is injected after the prematurely closed quoted string, causing the IMAP connection to be terminated.\n\n**More severe injection — exfiltrate emails from another folder:**\n```python\nsearch_emails(query='\" SEARCH RETURN (MIN) ALL')\n```\nProduces: `TEXT \"\" SEARCH RETURN (MIN) ALL\"` — injects a secondary SEARCH command.\n\n**Negative control — legitimate search:**\n```python\nsearch_emails(from_addr='user@example.com')\n```\nProduces: `FROM \"user@example.com\"` — correct, no injection.\n\n**Cleanup:** No persistent changes for read-only injection. For destructive injection (DELETE, EXPUNGE), impact persists.\n\n## Impact\n\nAn attacker who can craft prompts that cause an LLM agent to call `search_emails` with injection payloads can:\n\n- **Terminate IMAP connections** (denial of service)\n- **Inject arbitrary IMAP commands** — including LIST (enumerate folders), SELECT (switch folders), FETCH (read emails from other mailboxes), STORE (modify flags), COPY/MOVE (move emails), DELETE/EXPUNGE (permanently delete emails)\n- **Exfiltrate email contents** from folders the user did not intend to expose to the agent\n- **Permanently delete emails** via injected DELETE + EXPUNGE commands\n\nThe attack requires the IMAP backend to be configured (EMAIL_ADDRESS + EMAIL_PASSWORD env vars), which is a documented and common deployment for email-capable agents.\n\n## Suggested remediation\n\n1. **Escape double-quote characters** in IMAP parameters. Per RFC 3501, literal strings use `{n}\\r\\n` format or quoted strings with `\\` escaping:\n```python\ndef _escape_imap_string(s: str) -> str:\n    \"\"\"Escape a string for safe use in IMAP quoted strings.\"\"\"\n    # Use IMAP literal syntax for safety: {length}\\r\\n<data>\n    encoded = s.encode('utf-8')\n    return f'{{{len(encoded)}}}\\r\\n{encoded}'\n```\n\n2. Use IMAP literal syntax (`{n}\\r\\ndata`) instead of quoted strings for all user-controlled parameters. This prevents any injection regardless of content.\n\n3. Apply the escaping to all IMAP search criteria parameters: `from_addr`, `subject`, `query`, and `search_id`/`message_id`.",
  "severity" : [ {
    "type" : "CVSS_V3",
    "score" : "CVSS:3.1/AV:N/AC:L/PR:L/UI:N/S:U/C:H/I:H/A:N"
  } ],
  "affected" : [ {
    "package" : {
      "ecosystem" : "PyPI",
      "name" : "praisonaiagents"
    },
    "ranges" : [ {
      "type" : "ECOSYSTEM",
      "events" : [ {
        "introduced" : "0"
      }, {
        "fixed" : "1.6.59"
      } ]
    } ],
    "database_specific" : {
      "last_known_affected_version_range" : "<= 1.6.48"
    }
  } ],
  "references" : [ {
    "type" : "WEB",
    "url" : "https://github.com/MervinPraison/PraisonAI/security/advisories/GHSA-c969-5x3p-vq3v"
  }, {
    "type" : "PACKAGE",
    "url" : "https://github.com/MervinPraison/PraisonAI"
  } ],
  "database_specific" : {
    "cwe_ids" : [ "CWE-20", "CWE-77" ],
    "severity" : "HIGH",
    "github_reviewed" : true,
    "github_reviewed_at" : "2026-06-18T14:25:03Z",
    "nvd_published_at" : null
  }
}