Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type @ followed by the MCP server name and your instructions, e.g., "@ai-agent-logs-mcp list my recent sessions"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
MCP server so Cursor agents can talk to ai-agent-logs-api (same data shown in ai-agent-logs-webui).
Run the API (local http://127.0.0.1:8120 or published https://ai-agent-log-api.xaigrok.ir).
Node.js 18+.
Related MCP server: Log MCP Server
cd C:\Users\armin\GitHub\ai-agent-logs-mcp
copy .env.example .env
npm install
npm run build~/.cursor/mcp.json)"ai-agent-logs-mcp": {
"command": "node",
"args": ["C:/Users/armin/GitHub/ai-agent-logs-mcp/dist/index.js"],
"env": {
"AI_AGENT_LOGS_API_URL": "https://ai-agent-log-api.xaigrok.ir",
"AI_AGENT_LOGS_USERNAME": "armin",
"AI_AGENT_LOGS_PASSWORD": "dopadopa123"
}
}Restart MCP / reload Cursor after changing mcp.json.
Tool | API |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
log_agent_responseFull turn fields sent to the API: model, rate, date, time, device, user_ip, duration_ms, project, app (default cursor), token_usage, user_prompt, refined_prompt, is_accepted, agent_response, session, title_of_session.
Authenticated tools auto-login with env credentials when no JWT is cached.
Variable | Default |
|
|
|
|
|
|
Create a prompt log row (POST /api/v1/prompts). Auto-logs in if needed.
Inputs: mode (required), rate (required), title (required), device (required), +4 more.
Get one session and its prompt turns (GET /api/v1/sessions/:id). Auto-logs in if needed.
Inputs: id (required).
Check ai-agent-logs-api health (GET /health).
Inputs: no named parameters.
List stored prompts newest first (GET /api/v1/prompts). Auto-logs in if needed.
Inputs: no named parameters.
List agent conversation sessions newest first (GET /api/v1/sessions). Auto-logs in if needed.
Inputs: no named parameters.
Login to ai-agent-logs-api and cache the JWT. Optional username/password override env defaults.
Inputs: password, username.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
Scored across 6 tools
Each tool targets a distinct resource and action: health_check for API status, login for authentication, list/create for prompts, and list/get for sessions. No overlap or ambiguity among them.
All tool names follow a predictable lowercase_with_underscores pattern, mostly verb_noun (list_prompts, create_prompt, get_session). The singular 'login' and compound 'health_check' fit naturally without breaking consistency.
With only 6 tools, the set is tightly scoped for the purpose of a logs API. Each tool serves an essential function, and the count is well within the ideal range.
The surface covers core prompt and session operations (list/create for prompts, list/get for sessions) plus health and login. Missing a get_prompt or delete operations, but for a logging server these are often unnecessary, making the gaps minor.
Self-hosted AI prompt library: prompts, collections, tags, teams, chains. 29 MCP tools for agents.
Context and control for any AI. Agent identity, scoped access, and proof across tools.
Generate contextual prompts and reusable agent skills, evaluate prompts with the 16-dimension Prompt Score, and manage saved work in PromptDrive. Twelve MCP tools also provide authorized access to private Memory for source-grounded answers. Connect over Streamable HTTP using OAuth 2.1 and PKCE. Generation consumes account quota and automatically saves successful results; Memory access follows account permissions and plan limits.
Ingest and search LogsLoom logs from coding agents.
