agentsh
PulseGate's liveness check found it on 19 Sep 2026; it is registered on GitHub and has been in the index since 19 Sep 2026. How this is checked
agentsh is an open-source agent shell that intercepts file, network, and process activity at runtime. It applies deterministic policies to allow, deny, approve, or steer actions and records audit logs for AI agent deployments.
Inferred · not functionally tested
Overview
6 featuresPurpose: Preventing AI agents from performing unauthorized file, network, and process actions at runtime.
Inferred · not functionally tested
Audience: AI agent developers and security teams
Inferred · not functionally tested
Functions: monitoring
Inferred · not functionally tested
Interfaces: API: unknown · MCP: unknown · CLI: indicated (inferred, not tested) · Self-hosting: indicated (inferred, not tested)
Recorded constraints: pricing: open_source · license: Apache-2.0 · platforms: DESKTOP, CLI, WEB · deployment: browser, cli, macos, linux, self_hosted
Constraint provenance is unknown; confirm requirements with the publisher.
Record sources: agentsh.org · github.com. These links do not verify the individual claims.
In the AI & LLM security space, agentsh takes a focused approach. Inferred · not functionally tested: It focuses on preventing AI agents from performing unauthorized file, network, and process actions at runtime. Inferred · not functionally tested: It is built as an open-source project for AI agent developers and security teams. Basis unknown · not verified: The project is open source (Apache-2.0). Basis unknown · not verified: agentsh is available on the web, the command line, macOS, and Linux, and it can be self-hosted.
Behind agentsh is canyonroad, and it first shipped in 2025. The project is developed in the open on GitHub with 385 stars and 10 commits in the last 90 days. Inferred · not functionally tested: Key capabilities include runtime interception, policy enforcement, and file controls.
Summary written by a language model from the project’s public pages.
Tasks: Inferred · not functionally tested
- Runtime interception
- Policy enforcement
- File controls
- Network controls
- Process controls
- Audit logs
Topics: Inferred · not functionally tested
Built with & integrations
- Claude Code
- commit e11a32a007ac · since May 2026
- openai
- .openai in the HTML · bgpt- in the HTML
- anthropic
- bclaude in the HTML · bclaude- in the HTML
- Cloudflare
- cf-ray header · cf-cache-status header
- meta_llama
- llama- in the HTML
- google_gemini
- bgemini- in the HTML
Trust & compliance
Indexing history
1What PulseGate has recorded for this listing
- Indexed30 Aug · 08:18 UTCcanyonroad/agentsh seen via GitHub Search (Smart)Source: GitHub Search (Smart) · Open
Frequently asked questions about agentsh
- What does agentsh do?
- Inferred · not functionally tested: Agentsh focuses on preventing AI agents from performing unauthorized file, network, and process actions at runtime. It is catalogued under AI & LLM security on PulseGate.
- Who should use agentsh?
- Inferred · not functionally tested: agentsh is an open-source project built for AI agent developers and security teams.
- Is agentsh free?
- Basis unknown · not verified: Yes — agentsh is open source under the Apache-2.0 license and free to use.
- What platforms does agentsh run on?
- Basis unknown · not verified: agentsh runs on the web, the command line, macOS, and Linux. It can also be self-hosted.
- Is agentsh still maintained?
- PulseGate's liveness check found it on 19 Sep 2026. Its GitHub repository shows 10 commits in the last 90 days.
- What are alternatives to agentsh?
- Similar projects tracked by PulseGate include agentsecure, Agent Sentinel, and agentcn.agentsecureAgent Sentinelagentcn
- Who develops agentsh?
- agentsh is developed by canyonroad.
- How long has agentsh been around?
- agentsh first shipped in 2025.
Similar projects
Closest matches by what these projects do