analyze-ci-speed

name	analyze-ci-speed
description	Analyze compilation time and test durations from CI logs. Use when the user asks about slow builds, slow tests, or wants to optimize CI time.

Analyze CI Speed

Analyze compilation time and test durations from CI logs to identify optimization opportunities.

Prerequisites

• gh CLI installed and authenticated (gh auth login)

Instructions

Step 1: Get Recent Workflow Runs

Get latest successful run from dev branch:

gh run list --workflow=Rust --branch=dev --limit=1 --status=success --json databaseId,conclusion,headSha,createdAt

Step 2: Download Test Job Logs

Get test jobs (nextest, nextest_all_features, coverage):

gh run view <run-id> --json jobs --jq '.jobs[] | select(.name == "nextest" or .name == "nextest_all_features" or .name == "coverage") | {id: .databaseId, name: .name}'

Download logs, stripping ANSI codes:

mkdir -p ci-logs/speed-analysis/
gh run view --job <job-id> --log | perl -pe 's/\e\[[0-9;]*m//g' > ci-logs/speed-analysis/<job_name>.log

Step 3: Extract Compilation Time

Search for cargo build summary in logs:

Finished ... in Xm Ys

Also look for incremental compilation info and cache hit rates.

Step 4: Extract Test Durations

Parse nextest output for test durations. Look for patterns like:

PASS [   1.234s] crate_name::module::test_name
SLOW [  15.678s] crate_name::module::slow_test

Extract all test durations and sort to find top 10 slowest tests.

Step 5: Analyze and Report

Provide summary:

• Total compilation time
• Top 10 slowest tests with durations
• Any tests marked as SLOW by nextest
• Recommendations:
  • Tests that could be optimized or parallelized
  • Crates with long compilation times
  • Potential for better caching

Notes

• Run on successful builds to get complete timing data
• Compare across runs to identify regressions