Add info-level logging to all MCP tools (#250)

* Add info-level logging to all MCP tools and upload proxy

- Log entry params for browse_lists, use_list, progress, results,
  list_sessions, balance, and cancel (previously only logged on error)
- Log success outcomes (result counts, artifact_ids, balance amounts)
- Log upload_id + filename on presigned URL request
- Log upload_id + size on proxy start, artifact_id on proxy completion
- Log proxy error responses with status and body

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Use debug-level logging for progress polling, info only on terminal

Avoids log noise from the tight polling loop (~every 3s). Only logs
at INFO when the task reaches a terminal state (completed/failed/revoked).

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Log first and last progress poll, skip intermediate calls

Uses a module-level set to track which task_ids are being polled.
Logs at INFO on the first call ("polling started") and when the task
reaches a terminal state. Cleans up the set on terminal so re-polling
after a retry still logs correctly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Revert in-memory progress set, use debug+terminal INFO instead

The in-memory set doesn't work with multiple replicas — different
pods don't share state. Instead: log every poll at DEBUG level
(invisible at default INFO), log at INFO only on terminal state.
Task submission already logs when polling starts implicitly.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Fix tuple unpacking bug in everyrow_use_list

_fetch_task_result returns (df, session_id, artifact_id) but the call
site only unpacked 2 values, causing ValueError at runtime every time
everyrow_use_list was called.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

* Add logging to everyrow_list_session_tasks

New tool added on main was missing info-level logging.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

---------

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

Author: RafaelPo

everyrow-mcp/src/everyrow_mcp/tools.py

@@ -137,6 +137,11 @@ async def everyrow_browse_lists(
     Call with no parameters to see all available lists, or use search/category
     to narrow results.
     """
+    logger.info(
+        "everyrow_browse_lists: search=%s category=%s",
+        params.search,
+        params.category,
+    )
     client = _get_client(ctx)
 
     try:
@@ -156,6 +161,7 @@ async def everyrow_browse_lists(
             )
         ]
 
+    logger.info("everyrow_browse_lists: found %d list(s)", len(results))
     lines = [f"Found {len(results)} built-in list(s):\n"]
     for i, item in enumerate(results, 1):
         fields_str = ", ".join(item.fields) if item.fields else "(no fields listed)"
@@ -193,6 +199,7 @@ async def everyrow_use_list(
 
     The copy is a fast database operation (<1s) — no polling needed.
     """
+    logger.info("everyrow_use_list: artifact_id=%s", params.artifact_id)
     client = _get_client(ctx)
 
     try:
@@ -204,13 +211,18 @@ async def everyrow_use_list(
             )
 
             # Fetch the copied data and save as CSV
-            df, _ = await _fetch_task_result(client, str(result.task_id))
+            df, _, _ = await _fetch_task_result(client, str(result.task_id))
 
             csv_path = Path.cwd() / f"built-in-list-{result.artifact_id}.csv"
             df.to_csv(csv_path, index=False)
     except Exception as e:
         return [TextContent(type="text", text=f"Error importing built-in list: {e!r}")]
 
+    logger.info(
+        "everyrow_use_list: imported artifact_id=%s rows=%d",
+        result.artifact_id,
+        len(df),
+    )
     return [
         TextContent(
             type="text",
@@ -993,6 +1005,7 @@ async def everyrow_progress(
     unless the task is completed or failed. The tool handles pacing internally.
     Do not add commentary between progress calls, just call again immediately.
     """
+    logger.debug("everyrow_progress: task_id=%s", params.task_id)
     client = _get_client(ctx)
     task_id = params.task_id
 
@@ -1033,6 +1046,9 @@ async def everyrow_progress(
     ts = TaskState(status_response)
     ts.write_file(task_id)
 
+    if ts.is_terminal:
+        logger.info("everyrow_progress: task_id=%s status=%s", task_id, ts.status.value)
+
     return [TextContent(type="text", text=ts.progress_message(task_id))]
 
 
@@ -1044,6 +1060,7 @@ async def everyrow_results_stdio(
     Only call this after everyrow_progress reports status 'completed'.
     Pass output_path (ending in .csv) to save results as a local CSV file.
     """
+    logger.info("everyrow_results (stdio): task_id=%s", params.task_id)
     client = _get_client(ctx)
     task_id = params.task_id
 
@@ -1092,6 +1109,12 @@ async def everyrow_results_http(
     controls how many rows _you_ can read.
     After results load, tell the user how many rows you can see vs the total.
     """
+    logger.info(
+        "everyrow_results (http): task_id=%s offset=%s page_size=%s",
+        params.task_id,
+        params.offset,
+        params.page_size,
+    )
     client = _get_client(ctx)
     task_id = params.task_id
     mcp_server_url = ctx.request_context.lifespan_context.mcp_server_url
@@ -1186,6 +1209,11 @@ async def everyrow_list_sessions(
     Use this to find past sessions or check what's been run.
     Results are paginated — 25 sessions per page by default.
     """
+    logger.info(
+        "everyrow_list_sessions: offset=%s limit=%s",
+        params.offset,
+        params.limit,
+    )
     log_client_info(ctx, "everyrow_list_sessions")
     client = _get_client(ctx)
 
@@ -1251,6 +1279,7 @@ async def everyrow_balance(ctx: EveryRowContext) -> list[TextContent]:
     Returns the account balance in dollars. Use this to verify available
     credits before submitting tasks.
     """
+    logger.info("everyrow_balance: called")
     client = _get_client(ctx)
 
     try:
@@ -1266,6 +1295,7 @@ async def everyrow_balance(ctx: EveryRowContext) -> list[TextContent]:
             )
         ]
 
+    logger.info("everyrow_balance: $%.2f", response.current_balance_dollars)
     return [
         TextContent(
             type="text",
@@ -1293,6 +1323,7 @@ async def everyrow_list_session_tasks(
     Use this to find task IDs for a session so you can display previous results
     with mcp__display__show_task(task_id, label).
     """
+    logger.info("everyrow_list_session_tasks: session_id=%s", params.session_id)
     client = _get_client(ctx)
 
     try:
@@ -1341,6 +1372,7 @@ async def everyrow_cancel(
     params: CancelInput, ctx: EveryRowContext
 ) -> list[TextContent]:
     """Cancel a running everyrow task. Use when the user wants to stop a task that is currently processing."""
+    logger.info("everyrow_cancel: task_id=%s", params.task_id)
     log_client_info(ctx, "everyrow_cancel")
     client = _get_client(ctx)
     task_id = params.task_id

everyrow-mcp/src/everyrow_mcp/uploads.py

@@ -119,12 +119,13 @@ async def request_upload_url(
 
         try:
             engine_upload_url = data["upload_url"]
+            upload_id = data["upload_id"]
             # Rewrite the URL to point at the MCP server instead of the Engine.
             # The Claude.ai sandbox can reach the MCP server but not api.everyrow.ai.
             upload_url = _rewrite_upload_url(engine_upload_url, mcp_server_url)
             result = {
                 "upload_url": upload_url,
-                "upload_id": data["upload_id"],
+                "upload_id": upload_id,
                 "expires_in": data["expires_in"],
                 "max_size_bytes": data["max_size_bytes"],
                 "curl_command": f'curl -X PUT -H "Content-Type: text/csv" -T {shlex.quote(params.filename)} {shlex.quote(upload_url)}',
@@ -138,6 +139,12 @@ async def request_upload_url(
                 )
             ]
 
+        logger.info(
+            "Upload URL requested: upload_id=%s filename=%s expires_in=%s",
+            upload_id,
+            params.filename,
+            data.get("expires_in"),
+        )
         return [TextContent(type="text", text=json.dumps(result))]
 
 
@@ -205,19 +212,49 @@ async def proxy_upload(request: Request) -> Response:
         engine_url = f"{engine_url}?{request.url.query}"
 
     body = await request.body()
+    size_bytes = len(body)
     headers = {
         k: v
         for k, v in request.headers.items()
         if k.lower() in ("content-type", "content-length")
     }
 
+    logger.info(
+        "Upload proxy started: upload_id=%s size_bytes=%d",
+        upload_id,
+        size_bytes,
+    )
+
     try:
         async with httpx.AsyncClient(timeout=_PROXY_TIMEOUT) as http:
             resp = await http.put(engine_url, content=body, headers=headers)
     except httpx.HTTPError as exc:
-        logger.error("Upload proxy failed: %s", exc)
+        logger.error("Upload proxy failed: upload_id=%s error=%s", upload_id, exc)
         return JSONResponse({"detail": "Upload proxy error"}, status_code=502)
 
+    if resp.status_code >= 400:
+        logger.warning(
+            "Upload proxy error response: upload_id=%s status=%d body=%s",
+            upload_id,
+            resp.status_code,
+            resp.text[:200],
+        )
+    else:
+        # Parse artifact_id from Engine response for traceability
+        artifact_id = None
+        try:
+            resp_data = resp.json()
+            artifact_id = resp_data.get("artifact_id")
+        except Exception:
+            pass
+        logger.info(
+            "Upload proxy completed: upload_id=%s status=%d artifact_id=%s size_bytes=%d",
+            upload_id,
+            resp.status_code,
+            artifact_id,
+            size_bytes,
+        )
+
     return Response(
         content=resp.content,
         status_code=resp.status_code,