llms.txt
# Polylane
> Detects production issues and fixes them by opening pull requests.
Polylane is built by Coreplane Labs in San Francisco, founded by Boris Tane (built Baselime, acquired by Cloudflare; led observability at Cloudflare). Signing up is one command: `curl -fsSL https://polylane.com/setup | bash` installs the CLI, connects the coding agents on the machine, and creates the account. What the script does: https://polylane.com/setup. The tagline is "Nobody should be on-call."
## When to use Polylane
Reach for Polylane when the job is operating production software, not just observing it:
- **Detect production issues without threshold tuning.** Polylane agents read the actual metric, log, and trace series across AWS, Cloudflare, Vercel, Fly.io, Render, and Kubernetes, and flag regressions against each resource's own baseline.
- **Investigate an incident and prove the cause.** Every claim in a Polylane investigation links back to the query, log line, or change record behind it. Runs are shareable transcripts you can watch and interrupt.
- **Turn an investigation into a reviewed fix.** Investigations that end at a line of code end as a pull request with the full evidence trail. Human review and CI still gate the merge.
Polylane works with AWS, Cloudflare, Vercel, Google Cloud, Render, PlanetScale, Railway, Fly.io, Kubernetes, Supabase, Modal, Convex, Trigger.dev, ClickHouse, Turso, Datadog, Honeycomb, Axiom, Sentry, Better Stack, OpenStatus, Grafana Cloud, Logfire, GitHub, Devin, Cursor, Factory, Conductor, Capy, Slack, Linear, Mixpanel, PostHog, Depot, and MCP Server. Every integration is listed at https://polylane.com/integrations/. It is not a dashboarding tool or a metrics store: it sits on top of the telemetry you already have.
How an agent should call Polylane today: begin with https://polylane.com/auth.md to create or resume the user's account without a TTY, answer questions about the product from this file (or the MCP servers below), and once the user is signed in, drive the platform through the product MCP server at https://mcp.polylane.com/mcp.
## What Polylane does
Polylane runs a closed loop over production software, in five steps:
1. **Context**: agents build the context graph live, every resource and dependency across your clouds, observability providers and codebase, so every later step starts from real topology instead of guesses.
2. **Remediation**: agents fix the issue rather than the symptom. Polylane never changes production on its own: the fix is investigated first, written on a branch, and validated, then arrives as a pull request with the full evidence trail; your review and CI still gate the merge.
3. **Review**: agents review every pull request against the production it deploys to, whether an engineer or a coding agent opened it. Polylane walks the graph from the repository to the resources behind it and reads their tier, recent changes and open issues, then leaves one comment, updated in place on every push. A fail has to prove how a line in the diff breaks production, so passing is the default; turn on the blocking check and a fail holds the merge until a later push clears it.
4. **On-call**: agents answer the questions on-call gets, in the thread where they are asked. Who owns this, what changed, is it us: answered in Slack with every claim linking back to the query, log line, or change record behind it. Every run is a transcript you can watch, interrupt, and share.
5. **Coding agents**: your coding agent gets the same production context Polylane's own agents use, over one MCP endpoint with browser sign-in and OAuth handled for you: the context graph, telemetry queries, deploys and code search. Reads by default; writes sit behind a default-deny policy, and merging stays human.
The homepage walks the five steps as five reconstructions of the product at work, in the console, on a pull request, in Slack, and in a terminal, across three services. Context: the live graph around a Cloudflare Worker called checkout-edge. Remediation: the pull request Polylane opened, the diff that restores the pool, CI green, and Merge waiting for a person to press it. Review: the production-impact review Polylane left on a pull request in orders-api, with the write-rate chart behind the verdict. On-call: a question in a Slack channel answered straight from the graph, the follow-up answered with the series behind it, and a fix already on the way. Coding agents: a coding agent in a terminal, checking over Polylane's MCP endpoint who still calls POST /checkout before it changes the schema.
## Product capabilities
- **Context graph**: Nobody remembers how production is wired. Polylane keeps your code and cloud in one live graph, so your coding agent stops guessing too.
- **Issue detection**: No thresholds to tune. Polylane reads the charts and speaks up only when something is actually wrong.
- **Alert intelligence**: Forward your alerts. Polylane reads the query behind each one, tells you which are real, and keeps digging into those.
- **Impact intelligence**: Polylane reviews every pull request against the infrastructure it deploys to. The diff that would cause tonight's incident gets flagged before it merges.
- **Investigations**: Polylane works out the cause one step at a time. Every step shows its evidence, or admits it doesn't know.
- **Instrumentation**: You can't debug what you didn't log. Polylane finds the blind spots in your code and closes them.
- **Slack agent**: Ask about production in any channel and get an answer grounded in your real infrastructure.
- **Autofix**: Polylane writes the fix only once it has proven the cause, and checks it against the signal that raised the issue before you see it.
## Pages
- [Home](https://polylane.com/): the loop in five steps, with reconstructions of the product at work
- [Product](https://polylane.com/product): the platform capabilities
- [Context graph](https://polylane.com/product/context): A live graph of your cloud infrastructure: every resource and dependency across providers, code, and observability, kept fresh automatically. Your coding agent reads the same graph over CLI or MCP. (Markdown: https://polylane.com/product/context.md)
- [Issue detection](https://polylane.com/product/detection): AI issue detection with no thresholds to tune and no alert rules to maintain. Polylane reads your metrics, logs and traces, and tells you only when something is actually wrong. (Markdown: https://polylane.com/product/detection.md)
- [Alert intelligence](https://polylane.com/product/alert-intelligence): AI alert triage that cuts alert fatigue. Polylane checks every alert you forward, dismisses the noise with its reasons, and investigates the real ones. (Markdown: https://polylane.com/product/alert-intelligence.md)
- [Investigations](https://polylane.com/product/investigation): Automated root cause analysis. Polylane works out the cause step by step, and every step is backed by evidence or marked unknown. (Markdown: https://polylane.com/product/investigation.md)
- [Impact intelligence](https://polylane.com/product/impact-intelligence): AI pull request review against the live infrastructure each change deploys to, so the change that would break production gets caught before it merges. (Markdown: https://polylane.com/product/impact-intelligence.md)
- [Autofix](https://polylane.com/product/autofix): AI-written fixes for production issues. Polylane proves the cause first, then writes a fix for your team to review. (Markdown: https://polylane.com/product/autofix.md)
- [Instrumentation](https://polylane.com/product/instrumentation): Automated instrumentation: Polylane finds the observability gaps in your code and opens pull requests that close them, starting with the repository you connect first. (Markdown: https://polylane.com/product/instrumentation.md)
- [Slack agent](https://polylane.com/product/slack): Polylane in your Slack: mention it in any channel, DM it, or let it chime in when it has something useful to add. (Markdown: https://polylane.com/product/slack.md)
- [Use cases](https://polylane.com/use-cases): what teams hand to Polylane, one page per function (Markdown: https://polylane.com/use-cases.md)
- [Polylane for impact analysis](https://polylane.com/use-cases/impact-analysis): Catch issues before they break production: Polylane analyses the production impact of every pull request against the live infrastructure it deploys to. (Markdown: https://polylane.com/use-cases/impact-analysis.md)
- [Polylane for incident response](https://polylane.com/use-cases/incident-response): AI incident response: alerts triaged in seconds, automated root cause analysis that cites its evidence, and mitigation before the meeting starts. (Markdown: https://polylane.com/use-cases/incident-response.md)
- [Polylane for DevOps](https://polylane.com/use-cases/devops): AI for DevOps: Polylane keeps your cloud infrastructure in one live graph, tracks every change, and watches the queues and cron jobs nobody else does. (Markdown: https://polylane.com/use-cases/devops.md)
- [Polylane for ticket resolution](https://polylane.com/use-cases/ticket-resolution): AI ticket resolution: Polylane investigates each bug report with full production context, proves the cause, and writes the fix. (Markdown: https://polylane.com/use-cases/ticket-resolution.md)
- [For coding agents](https://polylane.com/for-agents): connect Claude Code, Cursor, or any MCP client to Polylane — setup prompt, the tools, unlimited reads (Markdown: https://polylane.com/for-agents.md)
- [Polylane for Claude Code](https://polylane.com/for/claude-code): Connect Claude Code to Polylane over MCP: query your clouds, telemetry, deployments and context graph without leaving the terminal. (Markdown: https://polylane.com/for/claude-code.md)
- [Polylane for Cursor](https://polylane.com/for/cursor): Connect Cursor to Polylane over MCP: real production context, clouds, telemetry and deployments, next to the code it writes. (Markdown: https://polylane.com/for/cursor.md)
- [Polylane for Codex](https://polylane.com/for/codex): Connect Codex to Polylane over MCP: it queries your telemetry, context graph and recent changes before it writes a line. (Markdown: https://polylane.com/for/codex.md)
- [Polylane for OpenCode](https://polylane.com/for/opencode): Connect OpenCode to Polylane over MCP: clouds, telemetry, deployments and the context graph, queryable from the terminal. (Markdown: https://polylane.com/for/opencode.md)
- [Polylane for VS Code](https://polylane.com/for/vscode): Connect VS Code to Polylane over MCP: agent mode queries your clouds, telemetry, deployments and context graph in the editor. (Markdown: https://polylane.com/for/vscode.md)
- [Polylane for Pi](https://polylane.com/for/pi): Connect the Pi coding agent to Polylane over MCP: clouds, telemetry, deployments and the context graph from your terminal. (Markdown: https://polylane.com/for/pi.md)
- [Pricing](https://polylane.com/pricing): every plan, price, and limit, rendered from the live billing API (Markdown: https://polylane.com/pricing.md · JSON: https://api.polylane.com/v1/public/billing/plans)
- [Security](https://polylane.com/security): what agents can touch, what they can't, and every gate in between (Markdown: https://polylane.com/security.md)
- [Integrations](https://polylane.com/integrations): every integration, with what Polylane does and how it connects; one page per integration at https://polylane.com/integrations/<type> (JSON: https://api.polylane.com/v1/public/integrations/catalog)
- [Blog — "Closed Loop"](https://polylane.com/blog)
- [Customers](https://polylane.com/customers/): the teams running production with Polylane
- [How Depot merged 53 Polylane fixes in its first month](https://polylane.com/customers/depot/): Depot uses Polylane across AWS, PlanetScale and ClickHouse to detect issues, find the root cause and open the fix, so the team reviews pull requests instead of chasing alerts.
- [How Supermemory ships fast without watching every deploy](https://polylane.com/customers/supermemory/): Supermemory was one of Polylane’s earliest customers, and uses it to watch every deploy across Cloudflare and PlanetScale, catch what breaks, and hand the team the cause and the fix.
- [How Terminal keeps hundreds of telematics integrations running](https://polylane.com/customers/terminal/): Terminal uses Polylane across AWS, Datadog and Better Stack to detect when something breaks, trace it to the cause and open the fix.
- [How Basedash merged 81% of Polylane's fixes in its first 24 days](https://polylane.com/customers/basedash/): Basedash runs on Kubernetes and uses Polylane to watch the paths its customers depend on, review every pull request against live production, and fix what breaks.
- [How Orchid makes sure its assistant never drops a message](https://polylane.com/customers/orchid/): Orchid uses Polylane to watch the message, email and calendar paths its users rely on, catch what breaks, and open the fix for review.
- [Careers](https://polylane.com/careers): open roles in San Francisco (Markdown: https://polylane.com/careers.md)
- [Services](https://polylane.com/services): the Polylane team takes your on-call and drives your alert volume toward zero (Markdown: https://polylane.com/services.md)
- [Polylane Developers](https://polylane.com/developers): developer portal — quickstart, API keys, OpenAPI spec, MCP servers, CLI, SDK, rate limits, versioning policy (Markdown: https://polylane.com/developers.md)
- [About](https://polylane.com/about): the company, the founder, why Polylane exists (Markdown: https://polylane.com/about.md)
- [Contact](https://polylane.com/contact): email, X, security reports (Markdown: https://polylane.com/contact.md)
## Blog posts
- [Nobody reads pull request descriptions, so we made videos](https://polylane.com/blog/nobody-reads-pull-request-descriptions-so-we-made-videos/): Almost none of the pull requests Polylane opens were asked for by an engineer, and a longer description does not make them easier to review. We are experimenting with a narrated video for every fix. Why, what it looks like, and what happened to merge rates. (Oct 14, 2026, by Vi Tran and Boris Tane. Markdown: https://polylane.com/blog/nobody-reads-pull-request-descriptions-so-we-made-videos.md)
- [Turn your app into a context graph](https://polylane.com/blog/turn-your-app-into-a-context-graph/): Agents that run software need a context graph of the app: every resource, what it connects to, the repository that deploys it and the team that owns it. How we built one on Durable Objects, keep it fresh across every provider without polling AWS, decide what counts as a change, and delete from it safely. (Oct 5, 2026, by Boris Tane. Markdown: https://polylane.com/blog/turn-your-app-into-a-context-graph.md)
- [We swapped our LLMs for Jev. It's 39% cheaper.](https://polylane.com/blog/we-swapped-our-llms-for-jev/): Our on-call agent makes thousands of small decisions a day. We moved them from LLMs to Jev, a model that doesn't write text, and cut the cost by 39%. (Sep 25, 2026, by Vi Tran and Boris Tane. Markdown: https://polylane.com/blog/we-swapped-our-llms-for-jev.md)
- [How we prevent slop from hitting prod](https://polylane.com/blog/how-we-prevent-slop-from-hitting-prod/): Nothing in your pipeline asks whether a change is safe for prod. This is the check we built that does, and how we got it to finish before your tests do. (Sep 21, 2026, by Vi Tran and Boris Tane. Markdown: https://polylane.com/blog/how-we-prevent-slop-from-hitting-prod.md)
- [Sub-agents are just wrong](https://polylane.com/blog/sub-agents-are-just-wrong/): Polylane's autofix used to be a workflow of multiple agents, each responsible for a single task, with an orchestrator managing state between them. We moved to a single agent which does the entire job, from evidence gathering, hypothesis testing to pull request. (Sep 14, 2026, by Aleksandr Diamond and Boris Tane. Markdown: https://polylane.com/blog/sub-agents-are-just-wrong.md)
- [How We Fixed Our Cloudflare Durable Objects Memory Exceeded Errors](https://polylane.com/blog/how-we-fixed-our-cloudflare-durable-objects-memory-exceeded-errors/): Cloudflare Durable Objects run in V8 isolates with a hard 128 MB limit. Ours idled at ~140 MB and was reset ~300 times a day. 130 MB of zod schemas were built at module load, most of them by code that was never called. How we profiled the production bundle, the two fixes, and the numbers, 218 MB to 82 MB and resets to zero. (Aug 26, 2026, by Aleksandr Diamond and Boris Tane. Markdown: https://polylane.com/blog/how-we-fixed-our-cloudflare-durable-objects-memory-exceeded-errors.md)
- [I'm Betting My Company on Proactive Agents](https://polylane.com/blog/proactive-agents/): Agents can do almost anything you ask. That's the problem, you still have to ask. The next evolution is agents that figure out the job to be done. I started with operations, because nobody should be on-call in 2026. (Jul 5, 2026, by Boris Tane. Markdown: https://polylane.com/blog/proactive-agents.md)
## Guides
- [How to keep AI coding agents from breaking production](https://polylane.com/learn/ai-in-production/how-to-keep-ai-coding-agents-from-breaking-production/): Stop AI coding agents breaking production: scoped credentials, branch protection they can't bypass, required checks that block, and production-aware review. (Markdown: https://polylane.com/learn/ai-in-production/how-to-keep-ai-coding-agents-from-breaking-production.md)
- [PlanetScale MCP Server: Setup and Production Permissions](https://polylane.com/learn/ai-in-production/planetscale-mcp-server/): Connect the official PlanetScale MCP Server to Claude Code, Cursor and Codex. Set narrow permissions, inspect Insights and review writes before execution. (Markdown: https://polylane.com/learn/ai-in-production/planetscale-mcp-server.md)
- [Cloudflare Containers agent sandboxes: startup and setup](https://polylane.com/learn/deployment-safety/cloudflare-containers-agent-sandboxes-startup-time/): Cloudflare Containers now start agent sandboxes in a median 648 ms. Set up the durable_object policy, runtime images and snapshots, and know the limits. (Markdown: https://polylane.com/learn/deployment-safety/cloudflare-containers-agent-sandboxes-startup-time.md)
- [How to correlate incidents with deploys using evidence](https://polylane.com/learn/deployment-safety/how-to-correlate-incidents-with-deploys/): Learn how to correlate incidents with deploys using release markers, metrics, logs and verified GitHub, Vercel, Cloudflare and Kubernetes change events. (Markdown: https://polylane.com/learn/deployment-safety/how-to-correlate-incidents-with-deploys.md)
- [What is change intelligence? A production safety guide](https://polylane.com/learn/deployment-safety/what-is-change-intelligence/): What is change intelligence? Learn how deploys, config, flags, infrastructure and dependency changes connect to incidents, using tools you already have. (Markdown: https://polylane.com/learn/deployment-safety/what-is-change-intelligence.md)
- [PlanetScale Insights: Debug MySQL and Postgres Queries](https://polylane.com/learn/monitoring-coverage/planetscale-insights/): Use PlanetScale Insights to find slow query patterns, anomalies and errors. Compare MySQL and Postgres features, test an index and close production alert gaps. (Markdown: https://polylane.com/learn/monitoring-coverage/planetscale-insights.md)
- [PlanetScale Best Practices: A Production DB Checklist](https://polylane.com/learn/reliability/planetscale-best-practices/): PlanetScale Best Practices for production: compare MySQL and Postgres branching, schema changes, connections, indexes, Insights, backups and engine limits. (Markdown: https://polylane.com/learn/reliability/planetscale-best-practices.md)
- [Supabase vs Convex in Production: Reliability and Cost](https://polylane.com/learn/reliability/supabase-vs-convex-in-production/): Supabase vs Convex in Pro
