smartcontractkit · kalverra · Jul 18, 2026
@@ -0,0 +1,86 @@
+---
+name: optimize-workflow
+description: Optimize CI runners and workflow structure for speed, simplicity, stability, and cost.
+disable-model-invocation: true
+---
+
+<initialization>
+Prompt user if missing:
+1. Workflow to optimize.
+2. Priorities (default: stability > speed > cost).
+3. "Speed" definition (e.g., cache hits vs raw execution).
+4. "Stability" definition (e.g., are flaky tests acceptable?).
+5. Spot setting (infer via scenarios, do not ask directly). [runs-on spot reference](https://runs-on.com/docs/costs/spot-pricing/)
+6. Acceptable structural changes (e.g., job splitting, caching, lint reorganization, trigger optimizations).
+
+Setup:
+1. Analyze workflow: identify jobs, runners, step dependencies, critical path, caching, and bottlenecks.
+2. Identify structural optimizations:
+   - Parallelize sequential steps/jobs (e.g., matrix builds).
+   - Add/fix caching (e.g., go modules, cargo, npm, docker layers).
+   - Move checks/lints before compilation/tests to fail early.
+   - Refactor triggers/scope.
+3. Permitted scope: High freedom. OK to introduce breaking edits to internal `smartcontractkit` actions/workflows to improve speed/simplicity.
+4. Ask user to target: specific job, whole workflow runner config, or workflow structure.
+5. Setup test workflow (bypass gates, mock inputs, add `workflow_dispatch`).
+6. Init/resume trial log: `.github/.agents/skills/optimize-workflow/trials/<workflow>.md`.
+7. Run baseline trial with current configuration to benchmark.
+</initialization>
+
+<constraints>
+- No OOM or Out of Disk Space failures allowed.
+- Prefer default `ubuntu-latest`. Use `runs-on` only for larger runner requirements.
+  - Note: `runs-on` with `tmpfs` enabled can cause silent OOMs (not visible in memory metrics) under high disk load. Toggle `tmpfs` off if unexplained OOMs occur, and evaluate speed/cost impact.
+- Verify runner configuration via available runners API before use.
+- Validate workflow YAML syntax (e.g., `actionlint` or dry-run) before committing.
+- Maintain semantic correctness. OK to break internal `smartcontractkit` action/workflow interfaces for speed/simplicity.
+- Use `gh` CLI for runs and PRs.
+- Change only one variable per trial.
+- Ensure fair comparison: cache status must be identical across trials compared for speed.
+- Document runner config and structure layout in each trial to ensure reproducibility.
+</constraints>
+
+<resources>
+- [runs-on docs](https://runs-on.com/docs/)
+- [available runners](https://go.runs-on.com/api)
+- [GitHub Action docs](https://docs.github.com/en/actions)
+</resources>
+
+<loop>
+1. Define trials. Update `<workflow>.md` log.
+2. User approves trials and execution method (parallel vs sequential).
+3. Push to new disposable branch.
+   - Commit message: `cpu=X/ram=Y` or `struct/[caching|parallel|ordering|simplify]`
+   - PR title: `[DO NOT MERGE] Trial: <workflow-name>`
+   - PR body: Details of configuration. Include before-after Mermaid diagram showing structural flow.
+4. Trigger workflow.
+5. Monitor run:
+   `python3 .github/.agents/skills/optimize-workflow/scripts/workflow_monitor.py [run_id] --format json --out-file .github/.agents/skills/optimize-workflow/trials/[trial-name].json`
+6. Compare trials:
+   `python3 .github/.agents/skills/optimize-workflow/scripts/workflow_compare.py .github/.agents/skills/optimize-workflow/trials/[trial-1].json .github/.agents/skills/optimize-workflow/trials/[trial-2].json --out-file .github/.agents/skills/optimize-workflow/trials/[trial-1]-[trial-2]-comparison.md`
+7. Update trial log. Include updated before-after diagram if structural changes occurred.
+8. Show condensed results. Prompt for more trials or stop.
+
+<trial-template>
+| Runner | Structure | Experiment | Expectation | Branch | Run ID | Commit | Stability | Runtime | Cost | Notes |
+|---|---|---|---|---|---|---|---|---|---|---|
+| `config` | `default / cached / parallel` | What's tested | Expected result | `branch` | `run-id` | `sha` | Pass/Fail/Flaky | mm:ss | $ | Findings |
+</trial-template>
+</loop>
+
+<complete trigger="User stops loop or all trials done">
+1. Recommend final configuration.
+2. Add PR summary table and before-after Mermaid diagrams:
+
+```md
+### [Workflow/Job Name] Runner & Structure Changes
+
+| Approach | Runner | Structure | Stability | Runtime | Runtime Delta (Abs/%) | Cost | Cost Delta (Abs/%) |
+|---|---|---|---|---|---|---|---|
+| [Old](https://github.com/link/to/baseline/workflow_run) | [original runner] | [original structure] | Pass/Fail/Flaky | mm:ss | +0:00 (+0%) | $ | +$ (+0%) |
+| [New](https://github.com/link/to/final/workflow_run) | [new runner] | [new structure] | Pass/Fail/Flaky | mm:ss | +0:00 (+0%) | $ | +$ (+0%) |
+```
+
+3. Clean debug lines. Make final approved edits on a new branch. Ask user to commit.
+4. Delete trial branches, logs, and PRs.
+</complete>
@@ -26,7 +26,7 @@ tools/clroot/db.sqlite3-wal
 .env*
 .dbenv
 !.github/actions/setup-postgres/.env
-/.github/.agents/skills/right-size-runners/trials/
+/.github/.agents/skills/**/trials/
 .direnv
 .idea
 .vscode/