Put eyes on the deploy
Picks the metrics that would show trouble, then watches them in Datadog and Sentry against baseline once the deploy lands.
I want to do this: We're deploying checkout-api at 5pm. Watch the metrics after and flag anything that moves. ## Setup (skip if Polylane is already set up) Read and follow https://polylane.com/auth.md for non-interactive signup and setup. Start by checking whether I am already signed in; reuse my account and workspace. If the CLI is missing, bootstrap it without starting the interactive wizard: curl -fsSL 'https://polylane.com/setup?ref=prompts' | bash -s -- --install-only Then follow the guide through email verification, workspace selection, source connections, and MCP authentication. Ask me for an email code or OAuth consent only when needed. Verify each step; report pending setup instead of claiming success from installation alone. ## How to work Over MCP: searchTools lists what this workspace exposes, with each tool's schema; call it first. runTool runs one tool, runCode chains several in one call and returns just the answer. search and execute cover the full Polylane REST API: threads, issues, investigations, autofixes, memories. From the terminal: the polylane CLI wraps the same API, with structured output and non-interactive flags everywhere. Reads always work. Write tools appear only if I have opted in, and every write is screened. ## Task: Put eyes on the deploy Steps: 1. Read the deploy's diff and work out which services and metrics it can affect 2. Record each metric's pre-deploy baseline 3. Compare the post-deploy window against baseline, rhythm-aware 4. Check error rates and logs for new shapes, not just volume 5. Deliver the verdict: clean, or exactly what moved with the series behind it Ground every claim in data you actually pulled: the query, the log line, the change record. If the data is inconclusive, say so. Ask me before anything that writes.
"Keep an eye on it" means nobody does.
The deploy lands, everyone watches the graph for four minutes, then dinner happens. The regression that takes twenty minutes to show up reports itself via customers.
- Four minutes of vigilance for a twenty-minute regression
- Nobody agreed which metrics even matter for this change
- The evening deploy that ruins the evening
One prompt, this much work. Every step on your real data.
- 1 Read the deploy's diff and work out which services and metrics it can affect
- 2 Record each metric's pre-deploy baseline
- 3 Compare the post-deploy window against baseline, rhythm-aware
- 4 Check error rates and logs for new shapes, not just volume
- 5 Deliver the verdict: clean, or exactly what moved with the series behind it
Vigilance that doesn't get bored
The right metrics identified from the diff and watched properly after the deploy. You hear about the regression from the watcher, not from customers.
The quiet failure 
“Find things that look wrong but never alerted: error spikes, dead crons, growing queues.” Know the account “Map our AWS account: what's running, what depends on what, and what's critical?” Risky config 
“Find the risky configuration: public buckets, missing alarms, single points of failure.” The cron audit “List every scheduled job across our clouds and when each one last succeeded.” Close the gaps 
“Which critical services have no logs or traces? Show the gaps and write the fixes.” Read my dashboards 
“Go through our dashboards and tell me what's drifting from normal this week.” Query to check “Write the query for checkout error rate by region over 24 hours, and save it as a check.” Coverage gaps “Which routes swallow errors or log nothing? Show the gaps and the fixes.” The weekly read 
“Summarise what our telemetry says about last week: regressions, improvements, anything odd.” Onboard me “I'm new here. Walk me through the architecture: the critical path, the data stores, what talks to what.” Instrument the launch “payments-v2 ships next week. What instrumentation should it have before launch? Draft it.”
“Find things that look wrong but never alerted: error spikes, dead crons, growing queues.” Know the account “Map our AWS account: what's running, what depends on what, and what's critical?” Risky config “Find the risky configuration: public buckets, missing alarms, single points of failure.” The cron audit “List every scheduled job across our clouds and when each one last succeeded.” Close the gaps “Which critical services have no logs or traces? Show the gaps and write the fixes.” Read my dashboards “Go through our dashboards and tell me what's drifting from normal this week.” Query to check “Write the query for checkout error rate by region over 24 hours, and save it as a check.” Coverage gaps “Which routes swallow errors or log nothing? Show the gaps and the fixes.” The weekly read “Summarise what our telemetry says about last week: regressions, improvements, anything odd.” Onboard me “I'm new here. Walk me through the architecture: the critical path, the data stores, what talks to what.” Instrument the launch “payments-v2 ships next week. What instrumentation should it have before launch? Draft it.”