PullStar — Better prepared 1-on-1s for engineering managers
P PullStar

For engineering managers

Walk into every 1-on-1 prepared.

PullStar turns GitHub activity into evidence-backed coaching briefs—so you can spot patterns, ask better questions, and have more useful conversations.

See a sample brief → Try the live demo

Open source · GitHub-first · Runs locally

PullStar 1-on-1 brief
Prepared for your next 1-on-1
Samantha Lee
Strong, consistent delivery. The clearest coaching opportunity is creating more leverage through review depth and smaller PR scope.
16
Pull requests mergedConsistent activity across all five weeks
8
Large pull requestsMore than 1,000 lines changed
20
Reviews completedNo written feedback detected
“How do you decide when a review needs detailed feedback rather than a quick approval?”
✓ No signup for the demo
✓ Evidence behind every recommendation
✓ Source code stays local

The problem

GitHub records the work. It does not prepare the conversation.

Managers are expected to remember weeks of delivery, reviews, and collaboration across an entire team. Important patterns disappear into the activity stream.

01 / ACTIVITY

Too much to reconstruct

Useful signals are scattered across pull requests, reviews, comments, repositories, and weeks of work.

02 / CONTEXT

Numbers without meaning

GitHub can show what happened, but not why a pattern may matter for coaching or team health.

03 / PREPARATION

The obvious gets attention

Urgent issues dominate while quieter opportunities for recognition, growth, and leverage are missed.

The difference

Good managers ask better questions.

PullStar does more than summarize activity. It connects evidence to a manager-relevant interpretation, then suggests a question worth asking.

Review depth Evidence: 20 approvals · 0 comments
“You are active in reviews, but most appear to be quick approvals. How do you decide when to leave detailed feedback?”
PR scope Evidence: 8 PRs over 1,000 lines
“Would smaller changes make it easier for the team to review, collaborate, and ship with confidence?”
Leverage Evidence: strong, consistent delivery
“Your delivery is consistently strong. Where would you like to create more leverage beyond your own output?”

Example output

A brief built for the conversation—not another dashboard.

Every interpretation stays connected to the supporting evidence, so the manager can apply judgment instead of accepting a black-box score.

1-on-1 Brief: Samantha Lee

High-confidence signals

Quick summary

Samantha shipped 16 pull requests in 30 days with consistent activity across all five weeks. Two patterns stand out: her pull requests are often large, and her reviews appear to be approvals without written feedback.

Supporting evidence

  • 16 pull requests merged, usually within one to two days
  • 8 pull requests exceeded 1,000 lines
  • 20 reviews given, with no written comments detected
  • Active in all 5 weeks of the analysis period
  • Review activity concentrated in one repository

What it may mean

Samantha appears to be a reliable, high-volume contributor. The opportunity is not performance correction; it is increasing her leverage through deeper review engagement and more reviewable PR scope.

Suggested conversation focus

“How do you decide when to leave detailed feedback rather than a quick approval?”

“Would smaller changes make your work easier for the team to review and collaborate on?”

“Where would you like to create more leverage beyond your own delivery?”

Manager note

Treat these as prompts, not conclusions. Team norms and project context may explain the observed patterns.

How it works

From engineering activity to a better 1-on-1.

PullStar turns raw activity into a short, manager-ready preparation workflow.

1

Collect evidence

Read pull requests, reviews, comments, timing, and collaboration signals from GitHub.

2

Identify patterns

Connect activity across time and interpret what may matter for coaching, recognition, or risk.

3

Prepare the conversation

Generate a concise brief with supporting evidence, context, and specific questions to ask.

Private by design

Your source code never leaves your machine.

PullStar runs locally. It converts repository activity into structured signals before AI processing. The model receives context and aggregated evidence—not your code.

1
GitHub activityRepository activity is read on your machine.
2
Local processingPullStar converts the activity into structured, aggregated signals.
3
AI interpretationThe model receives those signals and context—not source files.

Try PullStar

See whether it surfaces a question you would have missed.

Start with the live demo. Install locally when you want private repository access and full control.

Fastest path

Generate a sample brief

No login or setup. Enter a public GitHub username and see PullStar’s output in your browser.

Try the live demo →
Open source

Install locally

Use PullStar with private repositories while keeping control of your data and workflow.

git clone https://github.com/pullstar-ai/pullstar
View on GitHub
Agent workflow

Use the PullStar skill

Run PullStar within a supported agent workflow using the published PullStar 1-on-1 skill.

View the skill

Built in the open

Better preparation should lead to better conversations.

PullStar is early. Try it on a public GitHub profile and tell us what is useful, what is wrong, and what an engineering manager actually needs next.

Generate a brief Send feedback