LOCI GitHub Action

Line of Code Insights (LOCI) Agentic AI is the First Hardware-Aware Optimization Agent for CPUs. It predicts power spikes and Performance inefficiencies before test or inference. Optimizes code, configs, and serving - autonomously. It uncovers anomalies in static data, shifts observability left, and predicts emerging trends in dynamic data, discovering hardware and software reliability issues, to reduce costs and time to resolution.

LOCI lets you upload compiled binaries for performance analysis, view insights directly in your workflow summary, and (if you install the GitHub App) automatically receive PR comments with results.

• upload: uploads binaries in LOCI.

• summary: waits for the analysis to complete and attaches the Agent Report to the workflow run summary.

  Comments on PRs are posted automatically by the GitHub App after completed binaries upload.

Quick start

Before you begin:

• Add LOCI_BACKEND_URL as a GitHub Variable.
• Add LOCI_API_KEY as a GitHub Secret.
• Install the LOCI GitHub App on your repo if you want automatic PR comments and overall Performance Reviews.
• (optional, agentic mode only) If you want the summary step to add the summary report to the workflow job UI (note that the job will wait for the binaries upload and analysis to finish), add a Personal Access Token in a LOCI_GITHUB_TOKEN env variable for our action (we advice to store the actual token as a GitHub Secret).

Example: Build + Upload

name: LOCI Integration
on:
  workflow_dispatch:
  pull_request:
    types: [opened, synchronize, reopened]
  push:
    branches: [main]

env:
  LOCI_PROJECT: GHDemo

jobs:
  build-and-upload:
    runs-on: ubuntu-latest
    steps:
      - name: Prepare environment
        run: |
          sudo apt-get update
          sudo apt-get install -y gcc-aarch64-linux-gnu g++-aarch64-linux-gnu
        shell: bash

      - name: Checkout repository
        uses: actions/checkout@v4

      - name: Build sample
        run: |
          mkdir -p samples/build/bin
          aarch64-linux-gnu-g++ -o samples/build/bin/sample_01 samples/sample_01.cpp
        shell: bash

      - name: Upload Artifacts
        uses: auroralabs-loci/loci-dev-action@main
        env:
          LOCI_API_KEY: ${{ secrets.LOCI_API_KEY }}
          LOCI_BACKEND_URL: ${{ vars.LOCI_BACKEND_URL }}
        with:
          mode: upload
          project: ${{ env.LOCI_PROJECT }}
          binaries: samples/build/bin/

Example: Build + Upload + Summary (adds report and insights to job UI)

jobs:
  build-upload-and-summary:
    runs-on: ubuntu-latest
    steps:
      # ... build steps ...

      - name: LOCI Upload
        uses: auroralabs-loci/loci-dev-action@main
        env:
          LOCI_API_KEY: ${{ secrets.LOCI_API_KEY }}
          LOCI_BACKEND_URL: ${{ vars.LOCI_BACKEND_URL }}
        with:
          mode: upload
          project: ${{ env.LOCI_PROJECT }}
          binaries: samples/build/bin/

      - name: LOCI Summary
        uses: auroralabs-loci/loci-dev-action@main
        env:
          LOCI_API_KEY: ${{ secrets.LOCI_API_KEY }}
          LOCI_BACKEND_URL: ${{ vars.LOCI_BACKEND_URL }}
          LOCI_GITHUB_TOKEN: ${{ secrets.LOCI_GITHUB_TOKEN }} # required only for agent summary (if agentic mode is enabled)
        with:
          mode: summary
          project: ${{ env.LOCI_PROJECT }}

Good to know

• The upload step is enough for the GitHub App to add a comment on the PR (if installed and the company has agentic mode enabled).
• The summary step is only needed if you want the workflow run summary to show the LOCI Report and Top Function Insights.

Inputs

Outputs

LOCI Features Overview

How it works

• Upload step → always required; uploads binaries to LOCI backend.
• Summary step →
  • Standard mode → adds top-n Function Insights only.
  • Agentic mode → adds Agent Summary Report + Function Insights.
• GitHub App installed →
  • Posts LOCI Agent Summary Report as a PR comment (Agentic only).
  • Enables Performance Review Check Run (pass/fail).