Ship optimizations as PRs
Connect
SPIRITT
Polar SignalsPolar Signals
Performance fixes, found and shipped
SPIRITT reads your Polar Signals profiles like a performance engineer. It finds the functions and lines driving cost, maps who calls them and opens the pull request that fixes them.
What you can do with Polar Signals.
Hot paths found fast
Query function tables and flame graphs by profile type, labels and time window.
Down to the line
See per-line samples for a source file and the callers and callees around any function.
Hot paths turned into PRs
Pair profile evidence with your repository so optimizations ship as reviewed pull requests.
Works with your other apps.
Get started in three steps.
01
Open your workspace
Start a SPIRITT workspace for the services you want faster or cheaper to run.
02
Connect Polar Signals
Connect Polar Signals with access to your profiling projects, plus the repository that holds the code.
03
Delegate an outcome
Name the service, the cost or latency you care about and the bar to hit. SPIRITT finds the cause and ships the fix.
Hand over a mission.
Choose a mission. Make it yours.
Put Polar Signals to work.
Performance fixes, found and shipped
Questions
Polar Signals, with SPIRITT.
01What profiling data can SPIRITT query?+
Projects, profile types, labels and their values, function tables, flame graphs, caller and callee context, and per-line samples for a file.
02Does Polar Signals show my source code?+
It returns sample counts per line number. SPIRITT reads the matching file from your repository and lines the two up.
03Can it fix the code it finds?+
Yes. With your repository connected, SPIRITT implements the change and opens a pull request with the profile evidence.
04Which time ranges can it compare?+
Relative windows such as the last hour or day, or exact start and end times, so before-and-after comparisons stay precise.
05What about unsymbolized frames?+
SPIRITT can include them on request and flags services where missing symbols hide function names.
06Can it catch regressions after a deploy?+
Yes. It compares the same profile type before and after a release window and flags functions whose cumulative time grew, with flame graph context.