Two approaches based on your project's tooling:

Generated with /Users/mr.wolf/loadingalias/cargo-rail-testing/scripts/measure-impact.sh.

Interpretation:

• Task-runner repos (meilisearch, helix) show the largest immediate test/build skip savings.
• Standalone repos still gain deterministic targeting and explainable plan traces.

If you already have just, make, xtask, or shell scripts:

# cargo-rail provides change detection
PLAN=$(cargo rail plan --merge-base -f json)
SCOPE=$(echo "$PLAN" | jq -c '.scope')

# Your task runner handles execution
if echo "$PLAN" | jq -e '.surfaces.test.enabled' > /dev/null; then
  if echo "$SCOPE" | jq -e '.mode == "workspace"' > /dev/null; then
    cargo xtask test --workspace
  elif echo "$SCOPE" | jq -e '.mode == "crates"' > /dev/null; then
    mapfile -t CRATES < <(echo "$SCOPE" | jq -r '.crates[]')
    cargo xtask test "${CRATES[@]}"
  else
    echo "planner enabled test surface, but no package-scoped work was selected"
  fi
fi

Use .scope for package selection. Keep the full plan for gates, trace, and debugging. Do not rebuild execution scope from impact.

Why?

• cargo-rail stays focused on change detection
• Your existing build logic doesn't change
• Full control over execution
• No lock-in

Real examples: helix, meilisearch

If you don't have a task runner:

# cargo-rail handles both detection and execution
cargo rail run --merge-base --surface test
cargo rail run --workflow ci

Why?

• Single command for plan + execute
• No scripting required
• Built-in surfaces: build, test, bench, docs

Real examples: tokio, helix-db

Files that trigger full workspace rebuild:

[change-detection]
infrastructure = [
  ".github/**",       # CI changes
  "scripts/**",       # Build scripts
  "justfile",         # Task runner
  "deny.toml",        # License/security
  "rust-toolchain.toml",
  "Cargo.toml",
  "Cargo.lock",
]

[change-detection]
confidence_profile = "balanced"
bot_pr_confidence_profile = "strict"  # Override for dependabot, etc.

Detect non-Rust asset changes:

[change-detection.custom]
themes = ["runtime/themes/**"]
queries = ["runtime/queries/**"]
workloads = ["workloads/**"]

Plan output includes custom:themes, custom:queries, etc.

cargo rail config validate --strict
cargo rail plan --merge-base --explain
cargo rail run --merge-base --dry-run --print-cmd

• Configuration Reference
• Troubleshooting
• Validation Forks