- **CI routing-only edits, selected cheap core-test fixture edits, and narrow plugin contract helper/test-routing edits** use a fast Node-only manifest path: `preflight`, security, and a single `checks-fast-core` task. That path skips build artifacts, Node 22 compatibility, channel contracts, full core shards, bundled-plugin shards, and additional guard matrices when the change is limited to the routing or helper surfaces the fast task exercises directly.

- **Windows Node checks** are scoped to Windows-specific process/path wrappers, npm/pnpm/UI runner helpers, package manager config, and the CI workflow surfaces that execute that lane; unrelated source, plugin, install-smoke, and test-only changes stay on the Linux Node lanes.

The slowest Node test families are split or balanced so each job stays small without over-reserving runners: plugin contracts and channel contracts each run as two weighted Blacksmith-backed shards with the standard GitHub runner fallback, core unit fast/support lanes run separately, core runtime infra is split between state, process/config, shared, and three cron domain shards, auto-reply runs as balanced workers (with the reply subtree split into agent-runner, dispatch, and commands/state-routing shards), and agentic gateway/server configs are split across chat/auth/model/http-plugin/runtime/startup lanes instead of waiting on built artifacts. Broad browser, QA, media, and miscellaneous plugin tests use their dedicated Vitest configs instead of the shared plugin catch-all. Include-pattern shards record timing entries using the CI shard name, so `.artifacts/vitest-shard-timings.json` can distinguish a whole config from a filtered shard. `check-additional-*` keeps package-boundary compile/canary work together and separates runtime topology architecture from gateway watch coverage; the boundary guard list is striped into one prompt-heavy shard and one combined shard for the remaining guard stripes, each running selected independent guards concurrently and printing per-check timings. The expensive Codex happy-path prompt snapshot drift check runs as its own additional job for manual CI and for prompt-affecting changes only, so normal unrelated Node changes do not wait behind cold prompt snapshot generation and the boundary shards stay balanced while prompt drift is still pinned to the PR that caused it; the same flag skips prompt snapshot Vitest generation inside the built-artifact core support-boundary shard. Gateway watch, channel tests, and the core support-boundary shard run concurrently inside `build-artifacts` after `dist/` and `dist-runtime/` are already built.

The slowest Node test families are split or balanced so each job stays small without over-reserving runners: plugin contracts and channel contracts each run as two weighted Blacksmith-backed shards with the standard GitHub runner fallback, core unit fast/support lanes run separately, core runtime infra is split between state, process/config, shared, and three cron domain shards, auto-reply runs as balanced workers (with the reply subtree split into agent-runner, dispatch, and commands/state-routing shards), and agentic gateway/server configs are split across chat/auth/model/http-plugin/runtime/startup lanes instead of waiting on built artifacts. Normal CI then packs only isolated infra include-pattern shards into deterministic bundles of at most 64 test files, reducing the Node matrix without merging non-isolated command/cron, stateful agents-core, or gateway/server suites; heavy fixed suites stay on 8 vCPU while the bundled and lower-weight lanes use 4 vCPU. Broad browser, QA, media, and miscellaneous plugin tests use their dedicated Vitest configs instead of the shared plugin catch-all. Include-pattern shards record timing entries using the CI shard name, so `.artifacts/vitest-shard-timings.json` can distinguish a whole config from a filtered shard. `check-additional-*` keeps package-boundary compile/canary work together and separates runtime topology architecture from gateway watch coverage; the boundary guard list is striped into one prompt-heavy shard and one combined shard for the remaining guard stripes, each running selected independent guards concurrently and printing per-check timings. The expensive Codex happy-path prompt snapshot drift check runs as its own additional job for manual CI and for prompt-affecting changes only, so normal unrelated Node changes do not wait behind cold prompt snapshot generation and the boundary shards stay balanced while prompt drift is still pinned to the PR that caused it; the same flag skips prompt snapshot Vitest generation inside the built-artifact core support-boundary shard. Gateway watch, channel tests, and the core support-boundary shard run concurrently inside `build-artifacts` after `dist/` and `dist-runtime/` are already built.

Android CI runs both `testPlayDebugUnitTest` and `testThirdPartyDebugUnitTest` and then builds the Play debug APK. The third-party flavor has no separate source set or manifest; its unit-test lane still compiles the flavor with the SMS/call-log BuildConfig flags, while avoiding a duplicate debug APK packaging job on every Android-relevant push.

## Runners

| Runner | Jobs |

| -------------------------------- | ----------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

| `ubuntu-24.04` | Manual CI dispatch and non-canonical repository fallbacks, workflow-sanity, labeler, auto-response, docs workflows outside CI, and install-smoke preflight so the Blacksmith matrix can queue earlier |

| `blacksmith-4vcpu-ubuntu-2404` | `CodeQL Critical Quality`, `preflight`, `security-fast`, lower-weight extension shards, `checks-fast-core`, plugin/channel contract shards, `checks-node-compat-node22`, `check-guards`, `check-prod-types`, and `check-test-types` |

| `blacksmith-8vcpu-ubuntu-2404` | Linux Node test shards, bundled plugin test shards, `check-additional-*` shards, `check-dependencies`, and `android` |

| `blacksmith-16vcpu-ubuntu-2404` | `build-artifacts`, `check-lint` (CPU-sensitive enough that 8 vCPU cost more than they saved); install-smoke Docker builds (32-vCPU queue time cost more than it saved) |

| `blacksmith-16vcpu-windows-2025` | `checks-windows` |

| `blacksmith-6vcpu-macos-15` | `macos-node` on `openclaw/openclaw`; forks fall back to `macos-15` |

| `blacksmith-12vcpu-macos-26` | `macos-swift` on `openclaw/openclaw`; forks fall back to `macos-26` |

| Runner | Jobs |

| ------------------------------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- |

| `ubuntu-24.04` | Manual CI dispatch and non-canonical repository fallbacks, workflow-sanity, labeler, auto-response, docs workflows outside CI, and install-smoke preflight so the Blacksmith matrix can queue earlier |

| `blacksmith-4vcpu-ubuntu-2404` | `CodeQL Critical Quality`, `preflight`, `security-fast`, lower-weight extension shards, `checks-fast-core`, plugin/channel contract shards, most bundled/lower-weight Linux Node shards, `check-guards`, `check-prod-types`, `check-test-types`, selected `check-additional-*` shards, and `check-dependencies` |

| `blacksmith-8vcpu-ubuntu-2404` | Retained heavy Linux Node suites, boundary/extension-heavy `check-additional-*` shards, and `android` |

| `blacksmith-16vcpu-ubuntu-2404` | `build-artifacts`, `check-lint` (CPU-sensitive enough that 8 vCPU cost more than they saved); install-smoke Docker builds (32-vCPU queue time cost more than it saved) |

| `blacksmith-8vcpu-windows-2025` | `checks-windows` |

| `blacksmith-6vcpu-macos-15` | `macos-node` on `openclaw/openclaw`; forks fall back to `macos-15` |

| `blacksmith-12vcpu-macos-26` | `macos-swift` on `openclaw/openclaw`; forks fall back to `macos-26` |

Canonical-repo CI keeps Blacksmith as the default runner path for normal push and pull-request runs. `workflow_dispatch` and non-canonical repository runs use GitHub-hosted runners, but normal canonical runs do not currently probe Blacksmith queue health or automatically fall back to GitHub-hosted labels when Blacksmith is unavailable.