ci logs are a good starting point, but they're noisy and don't give you data you can filter or join across runs. structured test results let you ask which tests fail most, which are slow, and what's getting worse org-wide. now ga in depot. depot.dev/blog/now-ava...

Now available: Test results in your CI jobs

Report JUnit from your Depot CI or Depot GitHub Actions jobs to see failures in context. A single command runs your suite and uploads the report. Split suites by historical timing and they'll automati...

depot.dev

tar is sequential and gzip is stream-oriented, so to read one file in a container layer you unpack everything before it. soci adds metadata that lets the runtime seek directly to the bytes it needs. depot.dev/blog/why-soc...

Why SOCI belongs in the build

Most of a large container's startup goes to pulling and decompressing files it never reads to boot. SOCI makes the image seekable so the runtime loads only the bytes it needs, and Depot writes that in...

depot.dev

deleting from a content-addressable store is trickier than writing to one. you can't re-check s3 and delete in one step, so we record the version gc inspected and delete only that, staying safe even against a concurrent re-push. depot.dev/blog/tla-ver...

Improving system safety with Temporal Logic of Actions (TLA+)

Formal verification never seemed worth it: writing a faithful TLA+ spec took too long. But now agents can write the spec from your code. We tried it on the rebuilt Depot Registry garbage collector, an...

depot.dev

usage events tell you people are using your ai assistant. to see what they're trying to accomplish, you need traces: the user's question, the response, latency, session grouping, and where the conversation started. depot.dev/blog/using-p...

How we use PostHog to keep our AI assistant on track

Keeping our AI assistant working well is a human responsibility at Depot. PostHog's AI engineering features turned our never-ending reviews into 30 minutes a week.

depot.dev

test results are now ga in depot ci and github actions. ingest junit xml from your ci to get pass/fail/error/skip counts on every job, plus org-wide analytics for flaky, slow, and recurring failures. depot.dev/changelog/20...

Test results are now generally available | Changelog | Depot

depot.dev

kyle wrote up something we've been chewing on: the branch-pr-review-merge workflow was built for humans at human speed. agents blew past that, and github's collaboration model is now the bottleneck. depot.dev/blog/github-...

GitHub is the wrong shape for this new world

GitHub isn't just falling short on performance and reliability. The whole collaboration paradigm is the wrong shape for how we build software now.

depot.dev

best test for a process you inherited: ask what breaks if you delete it. "nothing, but people would be annoyed" means it's a ritual with no principle left. watts on keeping the why alive as you scale. depot.dev/blog/why-you...

Why your team processes suck

A team process without a principle underneath it is a real problem. Here's why the "why" is worth protecting as you scale.

depot.dev

we cut our slowest ci job from 6m 40s to 2m 36s. maxed vitest parallelism to use spare cpu (3m 54s), then sharded across 2 jobs with depot test splitting, now in beta. depot.dev/docs/ci/how-...

Split Depot CI tests with historical timing data

Split a test suite across matrix jobs with historical JUnit timing data, then run and report every shard.

depot.dev

fingerprint a failure too strictly and every run looks like a brand-new bug. too broadly and unrelated failures get lumped together. getting that balance right was the whole challenge, and we wrote up how we approached it. depot.dev/blog/failure...

Failure fingerprints: Comparing test failures over time

To make recurring test failures easier to reason about we needed to include historical context. Here's how we derive a stable identity for each test and a fingerprint for each failure, so you can tell...

depot.dev

the trick to getting consistent code out of an agent is constraint. cap cyclomatic complexity, function length, and file length. those guardrails force the agent to break work into small, readable units. depot.dev/blog/staying...

Staying in control of your codebase in the AI era

AI can produce code faster than you can review it, but you're still the one on call when it breaks. You need an AI-assisted workflow you trust: linter guardrails, reference tests the agent can copy, t...

depot.dev

tla+ always felt like too many weeks of modeling to win a prioritization fight, so we never budgeted it. now an agent writes the spec from our go, sql, and s3 calls while we make the judgment calls. tlc explored 14.3m states and caught a race we'd have shipped. depot.dev/blog/tla-ver...

Improving system safety with Temporal Logic of Actions (TLA+)

Formal verification never seemed worth it: writing a faithful TLA+ spec took too long. But now agents can write the spec from your code. We tried it on the rebuilt Depot Registry garbage collector, an...

depot.dev

Depot is now available in the Stripe Projects developer preview. Run 'stripe projects add depot', pick a plan, and get a Depot API token in your environment, ready for CI, runners, and fast container builds. depot.dev/blog/stripe-...

Depot is now available in Stripe Projects

Depot is joining the Stripe Projects developer preview. Developers and their agents can now set up Depot from the terminal: create or link an organization, pick a plan, and get a token back that unloc...

depot.dev

@planetscale.com's Vitess suite could fire 200–400 jobs from one PR. On a fixed runner fleet they queue up 20-30 min deep. A fixed fleet is the ceiling. Elastic capacity removes it, so there's nothing to queue behind. depot.dev/customers/pl...

How PlanetScale scrapped Buildkite and runs CI in half the time with Depot

How PlanetScale scrapped Buildkite and runs CI in half the time with Depot

depot.dev

keeping our ai assistant sherlock on track used to be endless manual toil: csv exports, a python script, a spreadsheet. we moved to posthog's ai observability and now weekly reviews take 30 minutes. depot.dev/blog/using-p...

How we use PostHog to keep our AI assistant on track

Keeping our AI assistant working well is a human responsibility at Depot. PostHog's AI engineering features turned our never-ending reviews into 30 minutes a week.

depot.dev

the same standup can feel like an interrogation or the most valuable 15 minutes of your day. the difference isn't the format, it's the principle underneath: control vs. connection. watts on protecting the why. depot.dev/blog/why-you...

Why your team processes suck

A team process without a principle underneath it is a real problem. Here's why the "why" is worth protecting as you scale.

depot.dev

Depot Code is now in private beta. It's the git host we wanted after three years accelerating everything downstream of source control: a diskless git server that stores content in S3 and scales horizontally. Regular git under the hood. depot.dev/blog/now-ava...

Now available: Depot Code in private beta

Source control has become the bottleneck at the top of the delivery loop. So we built our own: a diskless git server backed by blob storage that scales horizontally, mirrors your GitHub repos, and plu...

depot.dev

Depot CI and Sandbox workloads are now running on Depot Metal, our new 4th-gen compute platform. Benchmarks show them running 30% faster, with nothing to migrate. Built on bare-metal EC2 with microVMs and dedicated NVMe storage. depot.dev/blog/announc...

Announcing Depot Metal

Depot Metal is our next-generation compute and storage platform. Depot CI and Sandboxes are now running on it, with benchmarks showing workloads 30% faster and more optimizations on the way. You don't...

depot.dev

two test failures can look different and still represent the same failure pattern. the hard part isn't collecting failures. it's knowing when two failures are the same. iris shows how failure fingerprinting cuts noise and keeps the pieces that identify the pattern. depot.dev/blog/failure...

Failure fingerprints: Comparing test failures over time

To make recurring test failures easier to reason about we needed to include historical context. Here's how we derive a stable identity for each test and a fingerprint for each failure, so you can tell...

depot.dev

AI can generate code in seconds. That doesn't mean the hard part is solved. Someone still has to understand, review, maintain, and debug that code months later. Engineering guardrails like linting, reference tests, coding standards, and reviewable pull requests help make it easier to maintain.

New case study with @planetscale.com. PlanetScale runs CI at massive scale. After moving their CI workloads to Depot, they cut CI time in half on many workloads, eliminated runner management, and stopped thinking about CI. Read the full story: depot.dev/customers/pl...

How PlanetScale scrapped Buildkite and runs CI in half the time with Depot

How PlanetScale scrapped Buildkite and runs CI in half the time with Depot

depot.dev

Writing code stopped being the bottleneck. Validating it became one. When an agent ships a feature in 20 minutes, a 20-minute CI pipeline is the constraint. The code is done. The pipeline is what you're waiting on.

Your agent ships in seconds. Depot gives it unlimited concurrency, so nothing sits in a queue.

When your CI is agent-operable, a failed run is structured data, not a wall of logs. `depot ci diagnose --output json` returns the failed step and the error lines that matter. Your agent reads it, fixes the code, moves on.

1/ Paranoia used to be optional. AI-assisted bug hunting ended that. 🧵

macOS 26 (Tahoe) is now on Depot GitHub Actions runners. Change one line, `runs-on: depot-macos-26`, and you're on Apple M4 with Xcode 26 and the macOS 26 SDKs preinstalled. Simulators baked in too, no per-run downloads. depot.dev/blog/now-available-macos-26-github-actions

Reading CI logs is mostly scrolling past the 2,000 lines that don't matter to find the 4 that do. `depot ci diagnose --run <id>` Hands you the 4.

299,070 hours 👀 That's how many hours we've saved developers in the last week.

Albacore builds Ghostfin, a drone submarine. Its software ships as ARM64 containers, and on x86 that means QEMU. Their worst build ran overnight and still wasn't done. On Depot's native ARM64 builders, 30-40 min uncached builds now take 5-10. depot.dev/customers/al...

How Albacore builds ARM64 containers 6x faster with Depot

How Albacore builds ARM64 containers 6x faster with Depot

depot.dev

New in beta: test results for your CI jobs. Add depot/test-report-action after your test step, point it at your JUnit XML, and Depot shows pass/fail/error/skip on the job. Plus analytics for flaky, slow, and frequently-failing tests. Free in beta. depot.dev/changelog/20...

Test results for your CI jobs are now available in beta | Changelog | Depot

Depot can now ingest JUnit XML test reports from your CI jobs and show test results in the dashboard. Test results are free to try while in beta.

depot.dev

🚀 New in Depot CI: usage now breaks down per workflow on the org usage page. Attempts, vCPU minutes, elapsed time, avg duration, and error rate, plus health bars for the last 30 attempts. See which workflow is slow, flaky, or burning compute. depot.dev/changelog/20...