Backdoored code commits via poisoned repository prompts
Published May 31, 2026 · Updated May 31, 2026
Code injection in Aider 0.86.3 Architect Mode allows context-dependent attackers to commit backdoored code via poisoned repository text. ArchitectCoder passes prompt-influenced model output directly to editor_coder.run with preproc=False, causing the editor model to treat it as file-edit instructions. Exploitation requires Architect Mode to ingest attacker-controlled repository content; the demonstrated chain used --yes to modify auth.py, exfiltrate .env data, and commit the change.
Summary
What happened
Code injection in Aider 0.86.3 Architect Mode allows context-dependent attackers to commit backdoored code via poisoned repository text. ArchitectCoder passes prompt-influenced model output directly to editor_coder.run with preproc=False, causing the editor model to treat it as file-edit instructions. Exploitation requires Architect Mode to ingest attacker-controlled repository content; the demonstrated chain used --yes to modify auth.py, exfiltrate .env data, and commit the change.
The record
- CVE
- CVE-2026-10175
- Published
- May 31, 2026
- Updated
- May 31, 2026
- Vendor
- Aider-AI
- Product
- Aider
- Classifications
- CWE-74, CWE-94, T1195.002
- Attack vector
- network
- Privileges
- authenticated
Timeline
How it unfolded
- May 31, 2026CVE publishedPublication date reported by the CVE source.
- May 31, 2026Record updatedLatest update available in the CVE record.
Exploitability
Present is not the same as exploitable
Compare your product and version with the public record. A matching version still requires validation against your environment.
Is a vulnerable build present?
Compare these published version ranges with your installed build and any vendor patches.
- Affected versionversion=0.86.3
What conditions does exploitation require?
What is affected?
Attacks
What attackers are doing with it
Daily unique IPs observed by Shadowserver honeypots for known exploited vulnerabilities (KEVs). Missing observations do not establish an absence of attacks.
Weakness, pattern, technique
Public exploit references
- Architect mode can turn README prompt injection into committed backdoored codeproof of concept · demonstrated
Labels summarize the accepted research assessment. They do not indicate a test against your environment.
Technologies
Your stack
See the directory against your own environment.
Your stack
Check the software in your environment
Book a demo to see how Hinoki identifies affected software and validates exploitability in your environment.
Book a demo