Product-oriented plugins that give Codex, and future supported agentic harnesses, a consistent way to use Dell engineering services and governed development workflows.
| Plugin | Integration | Authentication |
|---|---|---|
dell-ai-factory-agentic-workspace |
Dell AI Factory Agentic Workspace; connector lifecycle and SSH | Interactive connector authentication |
dell-data-analytics-engine |
Customer-managed DDAE/Starburst MCP | Interactive OAuth |
dell-data-search-engine |
Customer-managed DDSE Kibana Agent Builder MCP | Least-privilege API key |
dell-data-orchestration-engine |
Local Dataloop Python SDK MCP server | Scoped Dataloop API key |
Live MCP tools/list is authoritative for the tools exposed by each gateway.
Write and destructive operations use Codex write approval mode.
codex plugin marketplace add https://github.com/dell/dell-harness-plugin.git
codex plugin add dell-ai-factory-agentic-workspace@dell-harness-plugin
codex plugin add dell-data-analytics-engine@dell-harness-plugin
codex plugin add dell-data-search-engine@dell-harness-plugin
codex plugin add dell-data-orchestration-engine@dell-harness-pluginSee each plugin README for its runtime prerequisites and authentication steps.
Each top-level plugin directory represents one Dell product or service. Shared workflow skills, documentation, evaluations, and support expectations remain in that product directory. Harness-specific connection files are adapters inside the same directory:
- Codex uses
.codex-plugin/plugin.jsonand.mcp.json. - Devin uses
.devin-plugin/plugin.jsonandmcp_config.jsonwhen that adapter has been implemented and validated.
Not every product supports every harness yet. The manifests present in each product directory are the authoritative statement of current support.
Dell AI Factory Agentic Workspace uses the Devspaces CLI connector for
authentication guidance, workspace status, start/stop/restart, and managed SSH
connections. Its plugin ID is dell-ai-factory-agentic-workspace (formerly
dell-devspaces-sandbox). To upgrade, run
codex plugin remove dell-devspaces-sandbox@dell-harness-plugin, then install
the new ID.
Provide Python with mcp and set DEVSPACES_CONNECTOR_COMMAND to the
connector executable. Set DEVSPACES_CLUSTER_URL to your Dev Spaces HTTPS
origin (required; there is no default) and, if the cluster uses a private
certificate authority, point NODE_EXTRA_CA_CERTS at that CA bundle.
Authenticate interactively with the connector before use.
Lifecycle writes and deletion remain gated. See dell-ai-factory-agentic-workspace/README.md for setup, connection lifetime, validation, and the optional legacy Kubernetes backend.
The DDAE, DDSE, and Dataloop plugins contain no customer endpoint or credential.
Set the environment variables documented in each plugin README in the
environment that starts the harness. DDAE uses interactive OAuth. Self-managed
DDSE uses a dedicated, least-privilege API key through
DDSE_MCP_AUTHORIZATION. Dataloop uses a scoped API key consumed by its local
Python SDK MCP server.
.agents/plugins/marketplace.jsonis the Codex marketplace catalog.- Harness-specific manifests and connection configuration stay inside the corresponding product directory.
requirements.tomlis an example Codex MCP allowlist for these plugins.- The repository contains no PAT values or other credentials, and examples use
reserved
example.com/example.testhostnames.
Contributions are welcome. See the contributors, and read
CONTRIBUTING.md before opening a pull request; every commit
must be signed off under the Developer Certificate of Origin (git commit -s).
Participation is governed by the Code of Conduct.
Do not report security vulnerabilities through public GitHub issues. Follow SECURITY.md.
Copyright (c) 2026 Dell Inc. or its subsidiaries. All Rights Reserved.
Licensed under the Apache License, Version 2.0. See NOTICE for attribution. Dell, Dell Technologies, and the Dell Technologies wordmark are trademarks of Dell Inc. or its subsidiaries; the license does not grant rights to use them except as required to describe the origin of the work.