CODING HARNESS MODEL LAYER

Use VAP Coding Plan API with Cursor, OpenCode, and Codex-style agents.

Cursor, OpenCode, Codex, and internal developer agents are harnesses. VAP supplies the model endpoint: an OpenAI-compatible Coding Plan API powered by Nemesis Deep Coder for long-context coding, review, planning, and agent workflows.

One model surface for multiple coding harnesses

Cursor: Use Coding Plan API where your Cursor workflow or adapter supports a custom OpenAI-compatible model endpoint.
OpenCode: Route terminal-first coding sessions to vap-code for planning, implementation support, and patch review.
Codex-style agents: Connect agent clients, repo automation, and CI-side coding workflows through the same /v1 model contract.

The contract

OpenAI-compatible by design

If your harness can call a custom OpenAI-compatible endpoint, point it at VAP and use the Coding Plan API key.

# Harness configuration values
Base URL: https://api.vapagent.com/v1
Model ID: vap-code
Auth: Bearer YOUR_CODING_PLAN_API_KEY

Direct API call

Use Chat Completions for broad compatibility, or the Responses endpoint for agent-style workflows documented in Developer Hub.

curl -X POST https://api.vapagent.com/v1/chat/completions \
  -H "Authorization: Bearer YOUR_CODING_PLAN_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"vap-code","messages":[{"role":"user","content":"Audit this feature branch and produce an implementation plan."}]}'

Powered by Nemesis Deep Coder

Coding Plan API gives these harnesses a flagship reasoning model with a 1 million-token context window, SOTA-level open-source coding performance, long-horizon mission capabilities, and stronger agent engineering.

FAQ

Is this a separate VAP product?

No. This is a Coding Plan API use case page. The product is Coding Plan API; Cursor, OpenCode, Codex, and internal agent tools are harness surfaces.

Do I need a Media API key?

No. Coding Plan API uses its own product-scoped key and plan access. Media API keys are for image, video, and music generation.

Does every harness support this natively?

Use native custom endpoint support where available. Otherwise, route through an adapter, proxy, or internal agent service that speaks OpenAI-compatible /v1.

Connect your coding harness to VAP.

Generate a Coding Plan API key and use vap-code in your agent workflow.

Get Coding Plan API Key View Coding Plans