Continue is the open-source AI coding assistant that runs in VS Code and the JetBrains IDEs. SearchAd AI's MCP server slots into its config the same way any other tool would. Run the launch campaign management from IntelliJ while your Swift code compiles. Same MCP server, same approval gate, same audit log.
mcpServers:
- name: searchad
type: streamable-http
url: https://searchad.ai/mcp
requestOptions:
headers:
Authorization: "Bearer sak_live_..."
Continue auto-detects changes. Open a chat, ask Continue what tools it has, and confirm searchad appears.
In the SearchAd dashboard at Settings → API access, generate a new key with a scope and daily envelope that match how aggressively you want Continue to act.
Add the searchad entry to mcpServers in ~/.continue/config.yaml. Or drop a per-workspace config.yaml inside the project for repo-scoped keys.
Open Continue chat in VS Code or your JetBrains IDE. Ask: "What apps do I have on Apple Search Ads?" If the tool resolves, you are live.
Most AI coding assistants pick a side: VS Code or JetBrains. Continue ships in both, with feature parity, so a team where the iOS lead uses AppCode and the backend lead uses VS Code can share the same Apple Search Ads workflow. Add SearchAd to Continue's config once and the integration travels everywhere the editor does.
The other reason teams pick Continue is the "bring your own model" property. SearchAd AI's MCP server does not care whether your Continue config is pointed at Claude, GPT-4o, Llama 3 via Ollama, or a custom endpoint. Tool calls travel over MCP regardless. The model affects the quality of the draft. It does not affect the safety guarantees.
"Show me the five Apple Search Ads campaigns with the highest CPT this week."
"Draft a Brazil launch with a $10/day budget and localized keywords. Paused."
"Give me a markdown summary of last week's ROAS by country and the top three changes I should consider."
"Which keywords in my Germany campaign converted in the last 14 days and what was the average CPT?"
Open your Continue config (~/.continue/config.yaml or the in-IDE config). Add an mcpServers entry named searchad with URL https://searchad.ai/mcp and an Authorization header with your bearer token. Continue picks up the change on next chat.
Yes. Continue ships extensions for both VS Code and the JetBrains family (IntelliJ, WebStorm, PyCharm, AppCode). The MCP integration is identical across them; same config, same tools.
Continue supports per-workspace config. Drop a config.yaml inside the repo with the MCP server scoped to that project. The global config still applies elsewhere; the workspace config takes precedence inside the project.
Continue lets you bring your own model (Claude, GPT, Llama via Ollama, others). The SearchAd MCP server works with all of them. The model affects draft quality; safety primitives do not change.
Continue persists chat history client-side; SearchAd's audit log persists every MCP call server-side. The two records match. Close Continue and reopen tomorrow; the audit log still shows what you did yesterday regardless of local chat state.