* fix(run): keep --last-failed reruns honest: failure-limit skips, carried debts, kept evidence
--last-failed now reruns tests --max-failures skipped. A test or failed
suite hook another filter leaves out stays owed in the report's new
run.carried until a rerun runs it; the github fold shows carried rows and
errors and stays red while anything is carried. A rerun prunes artifacts/
to what the previous report names and writes its attempts under
artifacts/rerun-<n>/ instead of wiping the tree and reusing attempt paths.
* fix(run): carry a setup's hook debt when no selected test needs it
The report has no row for an unneeded setup, so carryForward now asks the
collection whether a test still exists instead of the report's rows.
* fix(run): reject malformed artifacts in --last-failed reports, tighten rerun docs
Distinct ids in the carried fixture row.
* fix(run): keep carrying tests of a file a narrowed rerun could not collect
* fix(github): a carried interrupted test reads interrupted, not failed, in the folded page
* fix(report)!: count interrupted tests apart from failures
* fix(report): tell a failed test from its last failed attempt, whatever the cut retry left
* feat!: the app is declared on its target; engines only drive it
The app under test moves from the engine's options to `targets: [{ engine,
app }]`: `url`, `bundleId`, `appPath`, `identity`, `environment`,
`launchArguments`, `permissions`, `command`, and `readyUrl`. `web()` and
`mobile()` refuse every moved option in one `INVALID_CONFIG` naming the
target block; `defineEngine({ app })` is gone and engines check what their
platform needs in `validateApp`. `requires: ['browser' | 'native-app']` is
derived from the target's app. Replay cache and session keys are byte for
byte what they were, pinned by target-identity.test.ts.
Dependency services are not part of this version: `services` anywhere, and
`web({ services })`, fail saying so. `app.command` on port 0 without a
command, and `reuseExisting` on a free port, are refused. An `e2e mcp`
open cancelled or disconnected mid-start stops what it started, and a hung
shared start is abandoned once no session waits on it.
* fix: review follow-ups: cancelled opens, abandoned starts, null targets, stale docs
* refactor: drop the moved-option messages; old app keys are unknown keys
* test: give the replaced-nodes suite a budget that survives a loaded runner
* test: size the replaced-nodes hook budget to its seven tests
* docs: trim guides to what a test author needs
Cut runner internals, give each repeated topic one owner page, rewrite
the densest paragraphs.
Owners now:
- screenshots after a secret fill: security#screenshots-after-a-secret-fill
(auth, explore, debugging, agent-steps, tools, executors, engine link it)
- redaction rules and limits: security#redaction, #where-redaction-stops
- what the model sees/does in a step: agent-steps; security keeps the
trust-boundary and POLICY_DENIED tables
- budgets and timeouts: reference/agent#budgets (agent-steps keeps cost levers)
- assert under a custom executor, executor model conflicts: executors
- exit codes: debugging; telemetry in CI: telemetry; explore credentials:
explore; cache trust: security; personas: agents + authentication
- shard/last-failed edge cases: reference/cli; engine capabilities and
lifecycle detail: reference/engine
- locator strictness: reference/screen; no sessions on a device: mobile
Other changes:
- security: redactor internals -> short user-facing summary
- cache: timing internals cut, reason table one line per cell plus a
"Limits and gaps" subsection, committed-file changes as a list
- executors: tool-loop sample drives ctx.actions (portable, recorded)
- tools: simpler seed sample, mobileTools spread beside own tools
- writing-an-engine: locate in skeleton, minimal capability table, one-line
lifecycle, own Keyboard heading, tool-pack pointer, GitHub test link
- models: `ai` prerequisite and key source up front, env-var switch moved
- migrate: shared rules to one line + link, framework-specific "migrate
first", drop "Lives in" column, US spelling
- kernel: minimal setup sample, neutral replay size wording
- telemetry: source-checkout paragraph cut
* docs: address review on the guide trims
- tools: the seed sample reads APP_URL again
- ci: --last-failed reads report.json under output
- cypress, detox: strictness is for actions and single-node reads
- explore: artifacts, not evidence, land in the goal directory
* docs: fix stale and wrong claims in guides
- output: --artifacts -> --output in bug-bash guide and skill; sessions and
cache locations follow output / cache.dir in security; artifact path
gains the agent segment in github
- ci: upload artifacts on !cancelled() in ci guide and skill
- mcp: several sessions (--max-sessions) in security
- replay cache naming: "Commit the replay cache" heading and anchors
(old anchor kept, e2e init links the new one),
trace replay -> replay cache in writing-an-engine and executors
- timeouts: STEP_TIMEOUT / STEP_BUDGET_EXHAUSTED advice names config
timeout, judgmentTimeout, maxSteps, maxModelCalls (debugging, skill)
- mobile: app.open() in device examples and quickstart, mobileTools pack
all mutating, Android toolbar anchor, drop agent-device release note
- executors: verdict code rules, missing ctx members and scrollUntil,
a replay cache section; agent-steps passed+code rule
- migrate: swipe direction is the content's, `!` test exclusions,
video row status, Kernel for Selenium Grid
- secrets: an unset env string fails config load, a provider's value is
redacted only from its first fill; three secret sinks
- misc: Models link text, telemetry links outbound connections, integration
config wording, web-only expect(web) in a cross-platform block,
capabilities heading in engine reference, swipe direction in screen reference
* docs(skill): separate the credential redaction paragraph
* docs(reference): KERNEL_API_KEY names @e2e-dev/kernel