Native Swift · Built for agentic development · Nothing leaves your Mac
Loggerhead receives the logs, metrics, and traces from the software you run on this Mac, and shows them in one window. It is one Mac app, not a Docker stack, and nothing leaves your machine. It works with no internet connection. Telemetry linting then names the noise you would pay for in production.
Three signals, one window
Click a log line and jump to its trace. Click a span and see the logs it emitted. One local store keeps every signal in the same time base, so you stop stitching tabs together.
Severity tints, service colours, and filters for level and service. Search the message text, pause the stream, and open one line to read its fields.
Counters, gauges, and histograms land in DuckDB. Pick any window, group the series by label key, and compare services without an exporter in the middle.
A waterfall per request, spans grouped by service, and the slow one highlighted. Jump from a span to the logs it produced in one click.
Built for agentic development
AI coding tools guess when they cannot see. Loggerhead gives them a window into your running app through MCP, an open standard those tools already speak. The tool reads what your app actually did, then fixes the real problem.
WORKS WITH
>checkout is returning 500s locally. why?
loggerhead.query_logs { query: "checkout", severity: "ERROR", limit: 20 }
3 results · trace 7f3a…c1 in 2 of 3
loggerhead.get_trace { trace_id: "7f3a…c1" }
9 spans · payments → provider.charge 3,004 ms (timeout)
The 500 comes from payments. The call to provider.charge waits 3,004 ms and hits the 3 s client timeout in payments/client.go:41. The retry then double-charges. I will raise the timeout to 8 s and make the retry idempotent.
payments/client.go
- Timeout: 3 * time.Second,
+ Timeout: 8 * time.Second,
>█
One root, many branches
The app runs the receiver on localhost. Docker, Kubernetes, another Mac, or your app's SDK pushes into it as a branch. Loggerhead can also scrape a URL on a schedule, so tools that only expose metrics still show up. Every source gets an ID, so you always know where a record came from.
Loggerhead receiverruns inside the app · pushes in, pulls outListens on127.0.0.1 · port picked for you
collector.id = docker-mac
The app draws the same picture from live traffic. Every collector shows the services it carries, its record rate, and the moment it last spoke.
Telemetry linting
Most apps log too much and measure the wrong things. Loggerhead checks what your app emits in two ways. Lint rules are fixed checks that you set. AI analysis reads a longer window and finds the patterns a rule misses. Every finding comes with the fix already written. Copy it into the AI tool in your project, and the agent changes the code.
Lint rules
Forty-one rules read the last hour of logs, metrics, and spans every 30 seconds. They are deterministic: the same data gives the same findings each time. On the Rules tab of the Linting page, you turn each rule on or off and set its thresholds.
AI analysis
ProAI analysis sends one large prompt to a model and finds the problems that a fixed rule cannot see. You choose the window: the last hour, or the last 24 hours. It uses Apple Intelligence on your Mac first, with no key, and nothing leaves the Mac. For a stronger model, add your own AI provider key.
Made for the person who runs the app
Loggerhead is not a production observability platform, and does not want to be. It is the window you keep open next to your editor while you build, test, and stage.
One native app, one window, zero infra. See what your side project actually does before you ship it.
Point a staging cluster at your Mac over a tunnel and get a full picture of a release before it goes live. Staging and local stay apart: each environment keeps its own database.
Give Claude Code or Cursor the same view you have. The agent reads logs and traces over MCP, then acts on facts.
Honest scope: DuckDB on one Mac. Great for local and staging traffic. Keep your production platform for production.
Up in three steps
Download, open, and drag Loggerhead to Applications. The receiver, the store, and the MCP server all run inside the app. There is nothing else to install.
Inside Loggerhead, pick a source: an app on this Mac, Docker, Kubernetes, or a URL to scrape. Loggerhead picks a free port for you.
Pick the tool you use and click once. It can then read what your app did and fix real problems.
Works with any tool that supports MCP.
Buy once, keep it
$0forever
Everything a solo developer needs to see one app on one Mac.
$39one-time
For staging traffic, many collectors, and the alerting that catches a bad release.
$89one-time
Everything in Pro, and every future release. Pay once and never renew.
Questions
If your question is not here, email[email protected]and a person answers.
No. Loggerhead receives logs, metrics, and traces on 127.0.0.1 and writes them to a DuckDB file in your Application Support folder. There is no cloud account, no upload, and no third-party ingest. The one exception is your choice: if you pick a cloud model for AI analysis, the app sends a sample of your telemetry to that provider with your own key, after it masks values such as email addresses and IP addresses. With Apple Intelligence, the analysis stays on the Mac. The privacy page lists everything that reaches a server.
Yes. Loggerhead needs no internet connection. The receiver, the database, the lint rules, and the MCP server all run on your Mac. AI analysis with Apple Intelligence also runs on the Mac. Four things use the network: the check for app updates, on Pro a daily check that the licence is still active, AI analysis with a cloud provider key, and a problem report when you send one. The licence check sends the licence ID and nothing else. With no network, Pro stays on.
Loggerhead Free costs nothing and stays free. It covers one app on one Mac: logs, metrics, traces, the receiver, one branch collector, one pull source, the MCP server, and the 42 lint rules with the fix for each finding. Loggerhead Pro costs US$39 once, not per month and not per seat, and includes one year of updates. Loggerhead Lifetime costs US$89 once and includes every future release. Both add unlimited collectors and sources, unlimited data retention, AI analysis, and alerts. There is no ingest charge at any price.
You keep the version you have, and it keeps working. A Loggerhead Pro licence buys one year of updates. After that year you can renew for US$19 to get one more year of new versions, or you can stay on the last version the licence covered. A Lifetime licence gets every update and never needs a renewal, and a Pro licence upgrades to Lifetime for US$50. Loggerhead never stops working because a licence lapsed.
Loggerhead needs macOS 15 or later, and it runs on both Apple silicon and Intel Macs. It is one notarised Swift application with no runtime to install. There is no Linux build, no Windows build, and no hosted version.
You point your application’s OpenTelemetry exporter at http://127.0.0.1:4318 and give it a service name. If the application already emits OpenTelemetry data, that is a change to two environment variables. If it does not, you add the OpenTelemetry SDK for your language first. Loggerhead can also scrape a Prometheus metrics endpoint, which needs no change to the application at all.
Yes. Loggerhead hosts a Model Context Protocol server inside the app, on loopback. Claude Code, Cursor, Codex, Windsurf, and any other MCP client read logs, metrics, traces, and telemetry-quality findings through it. Nine tools only read. Seven tools change the lint rules or ignore a finding. Each of them needs a reason, and you can undo the change on the Rules tab of the Linting page. One tool marks a finding fixed after a code change. No tool deletes telemetry, and no tool reaches another machine.
Yes. Loggerhead runs one OpenTelemetry receiver on your Mac, the root. A standard OpenTelemetry Collector in Docker or in a Kubernetes cluster runs as a branch and forwards into that root. A container reaches the root at host.docker.internal. A cluster reaches it through a tunnel. Every record carries the ID of the collector that sent it.
No. Loggerhead is a development tool, not a production observability platform. It is the window you keep open next to your editor while you build, test, and stage a release. It holds local and staging traffic on one machine. It does not do on-call rotations, long retention across a fleet, or multi-user dashboards.
No. Loggerhead has no account and no sign-in. You download the disk image, drag the app to Applications, and open it. A Pro purchase sends a licence key to your email address, and you paste that key into the app.
Yes. Loggerhead keeps one workspace per project, and each workspace owns its own receiver port pair: 4317/4318 for the first, 4319/4320 for the second. Inside a workspace, each environment such as Local, Staging, or Production owns its own database, and a record picks its environment with the deployment.environment tag. The app shows one environment at a time, and you switch with one click in the sidebar. Free has one workspace with one environment; Pro has unlimited workspaces and environments.
Telemetry linting reads what your application emits and reports the problems before production charges you for them. It finds debug lines that repeat thousands of times an hour, labels with unbounded cardinality, an email address or a card number in a log field, an error that lost its trace context, and a metric that stopped arriving. Each finding carries a written fix that you paste into the AI coding tool that already has your project open. Free runs every lint rule and shows every fix.
The lint rules are deterministic checks, like a code linter: 42 rules read the last hour of telemetry, the same data gives the same findings, and you turn each rule on or off and set its thresholds on the Rules tab of the Linting page. AI analysis is a separate Pro check. It sends one large prompt to a model over a window you choose, such as the last hour or the last 24 hours, and finds patterns a rule cannot see, such as the same log line from two places or traces that do not line up. It uses Apple Intelligence on the Mac first, with no key, or a provider key that you add.
