run-evals
Skill22 starsChanged 3 months agoApache-2.0View on GitHub
Run the manzanas-eval scenario benchmark harness against a daemon and read its determinism report. Use when validating daemon changes end-to-end, measuring action latency, or checking reset determinism.
9 other projects use a version of this.
Adopted by
Most recently taken up by
- robertn702/reasoning-routertheir own version, first seen Nov 16, 2026
- stellar-experimental/stellar-raventheir own version, first seen Nov 15, 2026
- danielvm-git/bigpowerstheir own version, first seen Nov 6, 2026
- PyLops/pylops-agent-skillstheir own version, first seen Nov 1, 2026
- google/skills_lint.darttheir own version, first seen Oct 30, 2026
- lycorp-jp/sim-usetheir own version, first seen Oct 21, 2026
Other versions9 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 9
- lycorp-jp/sim-use★ 1.5ksame name
- danielvm-git/bigpowers★ 292same name
- Azure/azure-sdk-tools★ 134same name
- stellar-experimental/stellar-raven★ 17same name
- google/skills_lint.dart★ 8same name
- on-par/software-factory★ 6same name
- robertn702/reasoning-router★ 4same name
- PyLops/pylops-agent-skills★ 1same name
- rmatx/multi-agent-cs-crew★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository8 other files it gives its agents.
- broker-federationSkill
- build-and-testSkill
- drive-a-mirrored-iphoneSkill
- journal-evidence-exportSkill
- lease-and-drive-a-simSkill
- park-thaw-semanticsSkill
- run-the-daemonSkill
- ui-audit-and-predicatesSkill
The file
Files
Run the eval harness
eval/ is a pure protocol client: it runs YAML scenarios against any daemon
N times (fresh lease + reset per run) and reports pass/fail, determinism
rate, flaky steps, and per-step latency percentiles.
Build and run
go build -o bin/manzanas-eval ./eval/cmd/manzanas-eval
bin/manzanas-eval \
--daemon http://<mac-host>:7433 \
--runs 3 \
--out eval-out \
eval/scenarios/*.yamlOutputs: eval-out/report.md (human), report.json (machine),
<scenario>-<name>-run<N>.png screenshot artifacts. Exit code is non-zero on
any failing run or cross-run tree-hash drift, so it drops straight into CI.
Requirements & caveats (verified)
- The shipped scenarios need a real Mac daemon: they use
reset: eraseand dispatchlaunch_app/observe. Against a--mockdaemon the acquire fails with501 not_implemented: reset is not implemented in this build— that is expected, not a harness bug. For Linux/CI coverage of the harness itself usego test ./eval/...(httptest fake daemon). - Each run acquires a fresh lease with the scenario's reset policy, so eval
runs erase their target on release — point evals at a host/label where that
is acceptable, and prefer a device-specific label (e.g.
iphone-air) so the reset can't select another agent's simulator. - The runner renews the lease during long runs and deletes any snapshots the run created at teardown.
- Boots pass the daemon's safety gates; on a loaded host expect
503 overloadedretries to stretch run times.
Writing a scenario
YAML (strict parsing — unknown fields rejected; step names unique):
name: my-check
lease:
labels: [iphone-air] # device-specific label on shared hosts
reset: erase # none | erase | snapshot:<name>
ttl_seconds: 600
default_timeout: 120s
steps:
- op: boot
timeout: 300s
- op: action
kind: launch_app
payload: {bundle_id: com.apple.Preferences, terminate_running: true}
- op: assert
assert:
element_exists: {label: General}Step ops: boot, shutdown, action, fixture, snapshot, restore,
erase, wait, assert. Assertions (one per assert step, polled every 2s
until timeout): element_exists, element_absent (precede with a positive
assertion — it passes as soon as nothing matches), tree_hash
(save_as/equals_saved/equals), screenshot: {save: name}.
Shipped references: eval/scenarios/settings-navigation.yaml,
statusbar-fixture.yaml, snapshot-reset-determinism.yaml. Full docs:
docs/eval.md.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

