PCD MCP server
Commit Aug 27, 2026Provides a portable context layer that allows AI clients to access structured context, live work state, and accumulated knowledge across different platforms.
Also known as d3lanight/pcd, pcd
- Stars
- 0
- Tools
- —
- Downloads, 30 days
- —
- Latest version
- 0.4.0
- Last commit
- Aug 27, 2026
- License
- AGPL-3.0
What is PCD MCP?
PCD provides a portable context layer that allows AI clients to access structured context, live work state, and accumulated knowledge across different platforms.
Summary generated from the project's README
About PCD MCP server
PCD provides a portable context layer designed to carry structured information, live work states, and accumulated knowledge across different AI platforms. Instead of being tied to a single desktop application or IDE, users connect an MCP-capable client to a remote endpoint to access their personal containers. This setup allows for a consistent briefing of the AI regardless of which chat or development environment is being used.
Users access the service through a web dashboard to manage containers like skills, knowledge, and signals. Connection is established via streamable HTTP using OAuth 2.0 and PKCE authentication, meaning no manual API keys are required. Once connected, developers use commands such as load console to manage containers or load a specific container to enter a particular context. The system also includes experimental connections to link external tools through a single authenticated surface.
What you can do with it
- 1Sync work across environmentsA developer can move their active project state from a desktop chat interface to an IDE without losing progress.
- 2Brief an AI with domain expertiseAn AI client can be instantly prepared for a specific task using specialized context containers.
- 3Maintain long-term project findingsUsers can preserve important discoveries and insights that remain available after a session ends.
- 4Manage reusable AI capabilitiesA user can attach specific skills to a context deck to be used across different sessions.
Written from the project's README. Every use case is checked against a quote from it.
How to install PCD MCP
Formats verified 10 Nov 2026Add to CursorAdd to GooseInstall PCD MCP in Claude Code
$ claude mcp add --transport http pcd https://mcp.portablecontext.ai/mcpInstall PCD MCP in Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"pcd": {
"url": "https://mcp.portablecontext.ai/mcp"
}
}
}Install PCD MCP in VS Code
.vscode/mcp.json
{
"servers": {
"pcd": {
"type": "http",
"url": "https://mcp.portablecontext.ai/mcp"
}
}
}Install PCD MCP in VS Code CLI
$ code --add-mcp '{"name":"pcd","type":"http","url":"https://mcp.portablecontext.ai/mcp"}'Install PCD MCP in Codex
$ codex mcp add pcd --url https://mcp.portablecontext.ai/mcpInstall PCD MCP in Codex config.toml
~/.codex/config.toml
[mcp_servers.pcd] url = "https://mcp.portablecontext.ai/mcp"
Install PCD MCP in Gemini CLI
$ gemini mcp add -s user --transport http pcd https://mcp.portablecontext.ai/mcpInstall PCD MCP in Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"pcd": {
"url": "https://mcp.portablecontext.ai/mcp",
"type": "http"
}
}
}Automated checks
Ran Nov 15, 2026Full report →Not verified yet. A server needs a score of 80, no serious finding, and enough data for us to check. How checks work · What to fix for more points
- Source repository linked and reachabled3lanight/pcd is public and reachablegithub.com/d3lanight/pcd ↗Pass
- Package names this repository as its sourceNot checked
- Build provenance (Sigstore attestation)Not checked
- Open source license (OSI)AGPL-3.0spdx.org/licenses/AGPL-3.0.html ↗Pass
- Commit or release in the last yearLast commit or release 2026-08-27github.com/d3lanight/pcd ↗Pass
- OpenSSF ScorecardNot checked
- Package name not close to a popular packageNot checked
- No critical advisory in the package or its dependenciesNot checked
- No hidden instructions to the model0 tool descriptions and the README read; no override phrases, credential paths or hidden charactersPass
- Destructive tools are labelledNot checked
- Remote endpoint answers the MCP handshakeHTTPS endpoint answered 401 (sign-in required)Pass
README
View on GitHubPCD — Portable Context Deck
Your context. Any AI.
PCD is your portable context layer. Connect an MCP-capable client to one endpoint and it boots with your containers: structured context, live work state, and accumulated knowledge, composed by you and carried across every client you use. Use it in whichever AI you're working in—desktop, IDE, or chat—each already briefed, with nothing tied to a single client.
- Home: https://portablecontext.ai
- Dashboard: https://app.portablecontext.ai
- Docs: https://docs.portablecontext.ai
Access
PCD is in early access, with self-serve signup available in controlled batches. Go to app.portablecontext.ai to create an account when capacity is available.
If the current batch is full, you can leave your email there and we'll be in touch when capacity opens.
An account gives you the full product setup: an MCP endpoint for any compatible client, a web dashboard with a live view of your containers—skills, knowledge, signals, and work—plus connection management and export. The documentation is public.
Connect
Once you're in, point an MCP-capable client at the endpoint and sign in on first connect. No API key is required.
| Endpoint | https://mcp.portablecontext.ai/mcp |
| Transport | Streamable HTTP |
| Authentication | OAuth 2.0 + PKCE |
{
"mcpServers": {
"pcd": {
"url": "https://mcp.portablecontext.ai/mcp",
"transport": "http"
}
}
}
Then load console to create and manage your containers, load <container> to enter one, or load a specialist directly.
What you get
- Load — one call briefs your AI with your operating context.
- Specialists — specialized contexts scoped to a domain, role, or task.
- Work state — live work that follows you across clients.
- Knowledge — your accumulated, structured understanding.
- Signals — findings that outlive a session.
- Skills — reusable skills attached to a deck or shared across them.
- Connections (experimental) — connected external tools through one authenticated surface.
Data and privacy
Your data stays in your account—we don't sell it or train on it.
Source and licence
This repository contains PCD's public core under the GNU Affero General Public License v3.0 only. The mechanism is open and inspectable—AGPL guarantees it stays that way.
This source snapshot is not currently packaged as a runnable self-hosted distribution. A community self-hosted edition is on the table when demand justifies it. PCD's managed service, operational layer, and private product capabilities are not part of this repository.
The bundled fonts keep their upstream licences; see Third-party notices. PCD's names, logos, and brand assets are handled separately; see Trademarks.
Contributions
External code contributions are not currently accepted. Before opening contributions, PCD will publish a contributor licence agreement that explains the rights needed to maintain both the open-source and possible commercial licensing paths.
PCD-### references in code comments point to the project's internal issue tracker and are kept as provenance markers.
Built with PCD, inside PCD.
PCD MCP questions
How do I add PCD MCP to Claude Code?
Run claude mcp add --transport http pcd https://mcp.portablecontext.ai/mcp in your terminal. Add --scope user to make it available in every project.
How do I add PCD MCP to Cursor?
Open ~/.cursor/mcp.json and add the server under mcpServers, using the Cursor config shown in the install section above.
How do I connect PCD to my AI client?
Point an MCP-capable client to the provided endpoint and sign in on your first connection using OAuth 2.0.
Do I need an API key to use PCD?
No, an API key is not required for connection; it uses OAuth 2.0 + PKCE for authentication.
Is my data used for training?
No, your data remains in your account and is not sold or used for training.
Can I self-host PCD?
Currently, the repository is not packaged as a runnable self-hosted distribution, though a community edition is being considered for the future.