# CVE-2025-51482: Letta - RCE

> Live exploitation tracking for CVE-2025-51482 by the CrowdSec Live Exploit Tracker, built from real-world attack data crowdsourced from CrowdSec's global network.

Page: https://tracker.crowdsec.net/cves/CVE-2025-51482

## Key facts

- **Exploitation phase:** Limited Exploitation. The vulnerability is known but shows very limited attacker interest or exploitation activity.
- **CVSS score:** 8.8
- **Public exploit available:** Yes
- **Affected products:** letta-ai
- **Weaknesses:** CWE-94 (Improper Control of Generation of Code ('Code Injection'))
- **Tags:** ai_mcp

## Description

Remote Code Execution in letta.server.rest_api.routers.v1.tools.run_tool_from_source in letta-ai Letta 0.7.12 allows remote attackers to execute arbitrary Python code and system commands via crafted payloads to the /v1/tools/run endpoint, bypassing intended sandbox restrictions.

## CrowdSec analysis

[CVE-2025-51482](https://euvd.enisa.europa.eu/vulnerability/CVE-2025-51482) is a remote code execution vulnerability in letta-ai Letta 0.7.12, specifically within the /v1/tools/run endpoint, that allows attackers to execute arbitrary Python code and system commands by sending crafted payloads. This flaw enables threat actors to bypass intended sandbox restrictions, potentially leading to full system compromise and unauthorized access to sensitive data.

CrowdSec has been tracking this vulnerability and its exploits since 10th of September 2025.

Insights from the CrowdSec network reveal that the attackers trying to exploit CVE-2025-51482 are composed of a fairly even mix of opportunistic and targeted actors. Some attackers employ preliminary reconnaissance, while others use indiscriminate scanning.
CrowdSec data also reveals a clear uptick in attacks involving CVE-2025-51482 over the past week. Activity is above the usual baseline, suggesting growing attention from attackers. This may reflect rising awareness, recent exploit releases, or expanded targeting efforts.

Attackers exploit this vulnerability by sending crafted POST requests to the `/v1/tools/run` endpoint with a JSON payload containing malicious Python code in the `source_code` field, enabling remote command execution on the target server.

## Timeline

- 2025-07-22: CVE Published. CVE-2025-51482 is published to NVD.
- 2025-09-10: Rule Released. CrowdSec releases a rule to detect CVE-2025-51482 exploitation attempts against the CrowdSec Network.
- 2025-10-29: CrowdSec First Seen. CrowdSec observes in-the-wild exploitation of CVE-2025-51482 for the first time.

## References

- https://github.com/letta-ai/letta/pull/2630
- https://www.gecko.security/blog/cve-2025-51482
- https://github.com/letta-ai/letta
- https://github.com/XiaomingX/data-cve-poc/blob/main/2025/CVE-2025-51482/RCE.py

## Live exploitation data

The [Live Exploit Tracker page](https://tracker.crowdsec.net/cves/CVE-2025-51482) follows observed exploitation activity for CVE-2025-51482. [Request an API key](https://tracker.crowdsec.net/request-api-key) for full access to its exploitation timeline, attacking IP addresses and blocklists.