Performance Profiler¶

Identify hot paths, profile code, and propose meaningful performance improvements.

How it works¶

Detects the build system and benchmark infrastructure, runs existing benchmarks or instruments likely hot paths, and reports findings with concrete before/after measurements. The bar is high — files only when measurable profiling data supports the claim. Most runs end with noop.

Quick Install¶

mkdir -p .github/workflows && curl -sL \
  https://raw.githubusercontent.com/elastic/ai-github-actions/v0/gh-agent-workflows/performance-profiler/example.yml \
  -o .github/workflows/performance-profiler.yml

Trigger¶

Event	Schedule
`schedule`	Weekdays
`workflow_dispatch`	Manual

Inputs¶

Input	Description	Required	Default
`additional-instructions`	Repo-specific instructions appended to the agent prompt	No	`""`
`setup-commands`	Shell commands run before the agent starts	No	`""`
`allowed-bot-users`	Allowlisted bot actor usernames (comma-separated)	No	`github-actions[bot]`

Safe Outputs¶

create-issue — file a performance profiling report (max 1, auto-closes older reports)

Example Workflow¶

name: Performance Profiler
on:
  schedule:
    - cron: "0 14 * * 1-5"
  workflow_dispatch:

permissions:
  contents: read
  issues: write
  pull-requests: read

jobs:
  run:
    uses: elastic/ai-github-actions/.github/workflows/gh-aw-performance-profiler.lock.yml@v0
    secrets:
      COPILOT_GITHUB_TOKEN: ${{ secrets.COPILOT_GITHUB_TOKEN }}