sandbase-harness
A self-hosted AI-agent runtime with a native DeepSeek Harness MCP bundle for managing agents and sessions.
Install
$ dsh plugin --profile web add managed-agentsPlugin Overview & Capabilities
AI-assisted organization based on the public repository snapshot. The content must be grounded in source evidence and does not replace compatibility or security verification.
Key Capabilities
- Installs a DeepSeek Harness bundle that starts the managed-agents MCP bridge over stdio.
- Lets DSH list agents, create and run sessions, inspect results and artifacts, and stop work through native mcp__sandbase__ tools.
- Runs a local API and Console for managing agents, sessions, settings, and runtime state.
- Supports MCP toolsets, skills, credential vaults, permission policies, and approvals.
- Provides persistent SQLite-backed metadata, local workspace storage, audit trails, resumable event streams, and replay.
- Supports local-process, Docker, Kubernetes, and self-hosted-worker sandbox backends.
- Supports OpenAI, Anthropic, and OpenAI-compatible model providers.
Useful For
- Operate persistent AI-agent sessions from DeepSeek Harness.
- Run generated code in configurable local, Docker, Kubernetes, or worker-backed sandboxes.
- Inspect agent work through session events, audit trails, replay, results, and artifacts.
- Connect DSH workflows to a self-hosted agent runtime with MCP tools and permission controls.
Who It Fits
- DeepSeek Harness users integrating a self-hosted agent runtime.
- Developers building or operating persistent AI agents.
- Teams needing local infrastructure, session visibility, and configurable sandbox execution.
Documented Limitations
- Requires Node.js 22+ and npm 10+.
- Requires a model-provider API key for OpenAI, Anthropic, or an OpenAI-compatible endpoint.
- The DSH bundle expects a running managed-agents runtime, configured through MANAGED_AGENTS_URL.
- Docker is optional but required for Docker-backed sandboxes.
- Authentication is open by default until at least one API key exists.
DSH Compatibility
Version-specific runtime evidence collected by DSH Plugin. A missing result means we have not tested that combination yet.
Security Signals
Objective signals discovered from package metadata and source inspection. These are not a guarantee that a plugin is safe.
package.json declares 10 runtime, 13 development, 0 peer, and 0 optional dependencies.
package.json declares DSH bundle metadata.
GitHub reports the repository license as Apache-2.0.
A root package.json was captured and can be inspected by the registry.
Public GitHub source metadata is available for this registry snapshot.
Source & Registry Notes
Public provenance, Registry classification, and the latest source check for this entry, kept separate from runtime verification.
- Source repository
- sandbaseai/sandbase-harness
- Registry source
- GitHub · dsh-plugin topic
- Registry classification
- Plugin Bundle
- Source checked
- 9670796 · 2026-08-16
This project is independently indexed from public source information. DSH Plugin is not affiliated with DeepSeek or the plugin author. Always check the author repository before installation.
Repository Activity
- GitHub stars
- 593GitHub stars
- Forks
- 56
- Open issues
- 4
- Last commit
- 2026-08-15
- Last release
- 2026-08-15
Related DSH Plugins
Ranked by overlapping capabilities, use cases, plugin type, categories, and DSH profile.
DeepSeek Harness safety-gate and BlockRun LLM-provider plugin with review, vision configuration, and x402 wallet-paid model access.
View plugin →Clone-installable DeepSeek Harness bundle combining first-party plugins, bundled third-party plugins, and optional agent presets.
View plugin →Local-first persistent memory plugin for DeepSeek Harness, powered by Mnemon.
View plugin →Local document parsing and offline OCR tools for DeepSeek Harness.
View plugin →