AgentHound
PulseGate's liveness check found it on 14 Sep 2026; it is registered on GitHub and has been in the index since 31 Jul 2026. How this is checked
AgentHound is an open-source red team framework designed to test the security of modern AI agent infrastructure. It automates the full offensive lifecycle including reconnaissance, credential looting, prompt and model manipulation, and configuration attacks across MCP servers, A2A agents, inference platforms, vector stores, and related tools. Results are visualized as attack paths in a Neo4j graph, helping security teams understand and remediate risks in agentic systems.
Inferred · not functionally tested
Overview
5 featuresPurpose: Manually testing and mapping attack paths across complex AI agent infrastructure stacks.
Inferred · not functionally tested
Audience: red teamers
Inferred · not functionally tested
Functions: data_extraction, analytics
Inferred · not functionally tested
Interfaces: API: indicated (inferred, not tested) · MCP: unknown · CLI: indicated (inferred, not tested) · Self-hosting: indicated (inferred, not tested)
Recorded constraints: pricing: open_source · license: Apache-2.0 · platforms: API, CLI · deployment: cli, self_hosted
Constraint provenance is unknown; confirm requirements with the publisher.
Record sources: agenthound.io · github.com. These links do not verify the individual claims.
In the Penetration testing & red teaming space, AgentHound takes a focused approach. Inferred · not functionally tested: Manually testing and mapping attack paths across complex AI agent infrastructure stacks. Inferred · not functionally tested: It is built as an open-source project for red teamers. Basis unknown · not verified: The project is open source (Apache-2.0). Basis unknown · not verified: It ships for the command line, and it can be self-hosted.
It is developed by Adithyan AK, and it first shipped in 2026. The project is developed in the open on GitHub with 125 stars and 247 commits in the last 90 days. Inferred · not functionally tested: Key capabilities include Attack Path Mapping, Neo4j Graph Output, and Credential Looting. Inferred · not functionally tested: Catalogued interfaces include a public API.
Summary written by a language model from the project’s public pages.
Tasks: Inferred · not functionally tested
- Attack Path Mapping
- Neo4j Graph Output
- Credential Looting
- Prompt Poisoning
- Model Inversion
Topics: Inferred · not functionally tested
Built with & integrations
- local_oss
- bollama in the HTML · bvllm in the HTML
Trust & compliance
Indexing history
1What PulseGate has recorded for this listing
- Indexed31 Jul · 02:51 UTCadithyan-ak/AgentHound seen via GitHub Search (Smart)Source: GitHub Search (Smart) · Open
Frequently asked questions about AgentHound
- What does AgentHound do?
- Inferred · not functionally tested: Manually testing and mapping attack paths across complex AI agent infrastructure stacks. It is catalogued under Penetration testing & red teaming on PulseGate.
- Who is AgentHound for?
- Inferred · not functionally tested: AgentHound is an open-source project built for red teamers.
- Does AgentHound have a free plan?
- Basis unknown · not verified: Yes — AgentHound is open source under the Apache-2.0 license and free to use.
- What platforms does AgentHound run on?
- Basis unknown · not verified: AgentHound runs on the command line. It can also be self-hosted.
- Is AgentHound still maintained?
- PulseGate's liveness check found it on 14 Sep 2026. Its GitHub repository shows 247 commits in the last 90 days.
- What are alternatives to AgentHound?
- Similar projects tracked by PulseGate include agentic-redteam, agent-redteam, and Agentmetry.agentic-redteamagent-redteamAgentmetry
- Who makes AgentHound?
- AgentHound is developed by Adithyan AK.
- When did AgentHound launch?
- AgentHound first shipped in 2026.
Similar projects
Closest matches by what these projects do