bridge operator scripts (scripts/ quartet)
Census H22 — setup-runtime-bridge, bridge-health, verify-bridge, bridge-cleanup: the clone-side bash tooling that sets up, probes, smoke-tests, and cleans up after the aif bridge.
bridge operator scripts (scripts/ quartet)
Status: shipped-beta · Ships to: clone (operator tooling — packages/runtime-bridge/scripts/ is not part of the vendored consumer subset) · Fires at: run by the operator — setup at configuration time, health at session start, verify on demand ("did it actually work?"), cleanup when probe junk piles up.
What it is
Four bash scripts enumerated by ls packages/runtime-bridge/scripts/ → bridge-cleanup.sh bridge-health.sh setup-runtime-bridge.sh verify-bridge.sh (census enumeration, verified at pin). Two of them exist for the same documented reason (qloop-ux-probe 2026-06-01): "the costly, recurring bugs lived on the host↔container↔aif boundary that neither unit tests nor CI cover (a full agent run costs LLM tokens → no-paid-llm-in-ci). They were only ever found by a live probe MID-TASK" (scripts/bridge-health.sh:3-6).
How it works
setup-runtime-bridge.sh — "interactive consumer setup for the runtime bridge. Phase 1 scope: aif-handoff-or-skip (2-way). amux is Phase 2 (SW-H), not here" (scripts/setup-runtime-bridge.sh:2-4). "DETECT + INSTRUCT — never auto-installs aif-handoff": (1) "Probes whether an aif-handoff coordinator is reachable on RUNTIME_BRIDGE_AIF_URL"; (2) asks whether to enable the bridge (:5-9).
bridge-health.sh — "deterministic CONTAINER-side health check for the aif bridge … run it at session start (or after any bridge change) and get "bridge red, fix first"" (scripts/bridge-health.sh:2-7) — a $0 pre-flight for the boundary class unit tests can't cover.
verify-bridge.sh — "paste-and-run smoke test for the runtime bridge (REST dispatch) … the operator's "did it actually work?" check … It creates ONE throwaway task on your aif-handoff project and deletes it again" (scripts/verify-bridge.sh:2-6); exit/PASS-FAIL readable without an agent run.
bridge-cleanup.sh — "nuke the test/probe junk the bridge accrues each session … probe/smoke runs leave orphan aif tasks, /tmp ManualBackend kickoffs, and dedup-store cruft behind every session. This is the one-command sweep … It deletes ONLY clearly-test artefacts (title matches the test allowlist); real work is never touched" (scripts/bridge-cleanup.sh:2-6).
Fail mode / which channel fails: each script reports on its own stdout/stderr and exit code — health/verify are the FAILURE DETECTORS (red ⇒ fix before dispatch), cleanup is best-effort by design (test artefacts only), setup refuses to auto-install anything.
Satellites & companions
Health/verify exercise the same REST surface as the CLIs (H2, H11); cleanup sweeps the dedup log (H18) and ManualBackend /tmp artefacts (H17); the boundary bugs they probe are the ones the vitest suite cannot reach (H23).
Anchors
packages/runtime-bridge/scripts/setup-runtime-bridge.sh:2— «# setup-runtime-bridge.sh — interactive consumer setup for the runtime bridge.»packages/runtime-bridge/scripts/bridge-health.sh:2— «# bridge-health.sh — $0 deterministic CONTAINER-side health check for the aif bridge.»packages/runtime-bridge/scripts/verify-bridge.sh:2— «# verify-bridge.sh — paste-and-run smoke test for the runtime bridge (REST dispatch).»packages/runtime-bridge/scripts/bridge-cleanup.sh:2— «# bridge-cleanup.sh — nuke the test/probe junk the bridge accrues each session.»
runtime-bridge vendor copy (vendor/)
Census H21 — the S5 A7 vendored COPY (not a dependency): the minimum-for-dispatch subset of the bridge installed at --profile factory, with two deliberate omissions and an import-closed tree.
runtime-bridge vitest suite (test/)
Census H23 — the 23-file vitest suite covering dispatch, claim, park, answer, harvest, idempotency and the CLI entry plumbing, including the symlink-entry and import-no-sideeffect guards.