logsite for VS Code

See how often each log line fires in production, what it costs, and catch bad logging, right next to the logger.info(...) call.

logger.info("charging order %s for %d cents", order.id, total)   # 📈 1.2M in 24h · $42/mo
logger.debug(f"card {card_number} charged")                      # ⚠ LL001 eager f-string · LL002 PII
logger.warning("legacy discount path used for %s", order.id)     # no logs in 24h  ← dead code?

logsite maps production logs back to the exact line that wrote them, for Python (logging, structlog, loguru), Java (SLF4J, Log4j 2, java.util.logging, Lombok) and Go (slog, zap, zerolog, logrus, logr, klog), reading from Loki, OpenSearch / Elasticsearch, CloudWatch Logs or a log file.

Preview. JavaScript/TypeScript is planned.

Features

• Production counts above every log call, with a ▲ 8.0x marker when a line spiked.
• Cost per line: $/mo next to the count once you set your price per GB.
• Hover for the trend, the cost, and the last commits that changed the code around the call.
• Production errors where they're raised: the line that raised an exception gets TimeoutError 3k (new) in 24h, and the hover lists each error group with the log call that logged it.
• Log-quality lint as you type: eager formatting, secrets and personal data, lost tracebacks, failures logged at info, logs in hot loops, missing context like request_id.
• Quick fix that turns "order %s failed", order.id into a constant message with extra={"order_id": order.id} (or keyword arguments for structlog / loguru).

The terminal and CI get more: logsite errors (production errors grouped by cause), logsite trace <trace_id> (one request across services), logsite leaks, logsite cost, logsite spikes --since-deploy v1.4, and an MCP server for AI assistants. See the project README.

Setup

1. Install the server in your project's environment:

   pip install "logsite[lsp]"
   

   The extension uses the Python interpreter you've selected in VS Code (python -m logsite serve), then falls back to logsite on PATH. To use something else, set logsite.serverCommand. For Java and Go projects, which usually have no Python environment, install it once for your user: pipx install "logsite[lsp]".

2. Point it at your logs with a logsite.toml at the project root (lint works without one). In a monorepo, give each service its own; every file uses the nearest one:

   [backend]
   type = "loki"                  # or "opensearch", "elasticsearch", "cloudwatch", "file"
   url = "http://localhost:3100"
   selector = '{app="checkout"}'
   aggregate = true               # let Loki count per file/line: exact at any volume
   
   [cost]
   price_per_gb = 0.50            # optional: show $/mo
   

3. Add file and line to your log records for exact matching, e.g. python-json-logger's %(pathname)s %(lineno)d, structlog's CallsiteParameterAdder, or loguru's serialize=True. Without them, logsite matches by message template.

Commands

Settings

Develop

npm install && npm run compile
# press F5 in VS Code with this folder open to launch an Extension Development Host