TrojanForge Automated Repository Poisoning Against Coding Agents
TrojanForge
Researchers introduced TrojanForge, an automated framework designed to poison code repositories and compromise coding agents. The tool leverages abstract syntax trees and security review agents to deploy malicious payloads realistically.
- first report
- 19 Oct 2026
- last update
- 19 Oct 2026
Timeline · 1
Sources · 1
Research and catalogues
- 2026-10-19Trojan Horse: Automated Repository Poisoning against Coding AgentsarXiv · arxiv.org/abs/2610.23109
Titles and links belong to their publishers. The summary above is written by gcdRadar in its own words; data is CC BY-SA 4.0.
Classification
- lane
- Research
- AI role
- Attack on AI
- categories
- Data poisoningAgent hijack
- severity
- 1 · Research / proof of concept
- targets
- Enterprise deployment, Framework / tooling
- owasp llm top 10
- LLM04:2025 Data and Model PoisoningLLM06:2025 Excessive Agency
- mitre atlas
- AML.T0053 AI Agent Tool Invocation
- dates
- happened 19 Oct 2026 · first reported 19 Oct 2026
- origin
- News & research feeds
- record
- confidence 0.95 · created 2026-10-19 05:07Z · updated 2026-11-03 03:04Z
Lane, AI role, categories and severity come from the classifier; the OWASP and ATLAS rows are derived from the categories. How these are assigned
Copy a brief
This record as a short Markdown brief, for a ticket, a chat or a coding agent.
the brief as text
# TrojanForge Automated Repository Poisoning Against Coding Agents https://gcdradar.vercel.app/incidents/trojanforge-automated-repository-poisoning-against-coding-ag · AIR-2026-1394 · closed 2026-11-02 - Lane: Research - Severity: 1 (Research / proof of concept) - Fix status: Fix status unknown - First reported: 2026-10-19 - AI role: Attack on AI - Products: TrojanForge ## Summary Researchers introduced TrojanForge, an automated framework designed to poison code repositories and compromise coding agents. The tool leverages abstract syntax trees and security review agents to deploy malicious payloads realistically. ## What to do No action recorded; read the sources below and the vendor's advisory. ## Latest updates - 2026-10-19: Trojan Horse: Automated Repository Poisoning against Coding Agents (https://arxiv.org/abs/2610.23109) ## Sources - 2026-10-19 arXiv: Trojan Horse: Automated Repository Poisoning against Coding Agents https://arxiv.org/abs/2610.23109 Verify against the vendor's advisory before changing production. Data CC BY-SA 4.0, https://gcdradar.vercel.app.
Related
- liveLLM Agent When Experience Becomes a Backdoor: Poisoning Self-Evolving LLM Agents via Failure-Aware Task Optimization
Research · since 10 Nov 2026
AI - liveLLM Agent Memory Managers Accept Changes, Reject Attacks: Robust Memory Management for LLM Agents
Research · since 2 Nov 2026
AI - liveMobileProbe: Adaptive Security Evaluation of Mobile Agents via Trajectory-Aware Injection
Research · since 13 Nov 2026
AI - livePrompt Injection, Esq.: pwning AI agent guardrails with lawyerspeak
Research · since 11 Nov 2026
AI - liveAI-assisted cyberattack breaches German web host assistant interface
Breaches & attacks · Exploited · since 9 Nov 2026
AI - liveLLM agents Defending LLM Agents Against Parameter Poisoning
Research · since 4 Nov 2026
AI