hindsight-copilot

Long-term memory for GitHub Copilot in VS Code, powered by Hindsight.

hindsight-copilot init wires the Hindsight MCP server into VS Code's .vscode/mcp.json and adds a recall/retain rule to .github/copilot-instructions.md. Copilot's agent mode then has recall / retain / reflect tools and — guided by the rule — recalls relevant memory at the start of a task and retains durable facts as it works.

How it works

VS Code Copilot supports two things this integration uses:

• MCP servers in .vscode/mcp.json (agent mode), including HTTP servers with headers — so the Hindsight MCP endpoint connects directly:

  {
    "servers": {
      "hindsight": {
        "type": "http",
        "url": "https://api.hindsight.vectorize.io/mcp/my-project/",
        "headers": { "Authorization": "Bearer hsk_..." }
      }
    }
  }
  

• .github/copilot-instructions.md, which Copilot applies to every chat in the workspace — that's where the recall/retain rule lives.

Install

pip install hindsight-copilot
cd your-project
hindsight-copilot init --api-token YOUR_HINDSIGHT_API_KEY --bank-id my-project

init merges the servers entry into ./.vscode/mcp.json and writes the rule into ./.github/copilot-instructions.md. Reload VS Code, open Copilot Chat in agent mode, and start the hindsight MCP server from the chat's tools menu.

Use a Hindsight Cloud key, or a self-hosted server with --api-url http://localhost:8888 (no token needed for an open local server). If mcp.json has comments, init prints the snippet to paste instead of touching the file — or run hindsight-copilot init --print-only anytime.

Commands

Configuration

Development

uv sync
uv run pytest tests -v -m 'not requires_real_llm'   # deterministic suite
uv run pytest tests -v -m requires_real_llm          # gated MCP-endpoint check

License

MIT