- stateVersion starts at 1: the package is new, no persisted rows predate it
- pendingCalls pairs by own key so a provider callId naming a prototype
property cannot fold toolMs to NaN on an unmatched crash-recovery result
- StatsLine folds the window fallback only when no sessionStats value is
served, and gates the token group on actual token activity instead of
steps, so failed-only sessions drop the zero-token group
- correct the crash-step counting semantics in the README and Agent Note:
recovery closes interrupted steps with a synthetic step/end on reload
- reword the window-scoped alternative as a plain rejected option and name
isTokenDelta's home beside the StreamChunk type
cancel-in-progress is evaluated against the newly triggered run, so
exempting push means one master merge does not cancel the drill still
running from the previous one — not that a drill always finishes. A
benchmark dispatched on master shares the group and does cancel a
mid-flight drill. Record that bound in the runbook and drop the
overstated wording from the workflow comment and the spec name.
Also repair two gates that fail on master and block every pull request:
the telemetry note referenced an SDK proposal deleted in 408721954a,
and the ui-settings-general README pair carried stale recorded hashes
after both sides were updated together in aa1ec02bc6.
The two self-hosted standby drills each run their complete unsharded
aggregate with one gate worker, which takes longer than the interval
between master merges, so unconditional cancel-in-progress supersedes a
drill before it reaches a verdict and the lane yields no readiness
evidence for the failover runbook to point a responder at.
Exempt push and nothing else. This has to be decided at workflow level:
cancellation applies to the whole superseded run, so a job-level
concurrency group cannot exempt its job. The negated form is
load-bearing — naming pull_request alone would also stop cancelling
workflow_dispatch, and each runner benchmark fans out to twelve larger
runners for up to fifteen minutes in this same group on master, so a
re-dispatch would queue ahead of a drill instead of replacing a stale
measurement. It does not promise every push run finishes: a newer
pending run still displaces an older one, only that the lanes
periodically reach a verdict.
A master push carries only wine-apt-cache and the two drills; every other
job is pull-request-gated, workflow_dispatch-gated, or if: false. The
spec pins that set and classifies by exact condition, since a negated
event test mentions the event it excludes.
Keep both Remote contributions master and this branch add: the mount loop
now carries commandsRemote, goalsRemote, pluginInventoryRemote, and
messageFeedbackRemote, with both new tsconfig references retained.
The declared-provider-catalog note still recorded modalities as closed
because "nothing reads them: context.ts keeps only text blocks" — the
same reasoning this change disproves. Rewrite that fact in place, keep
pricing closed for its own still-current reason, and cross-link both
directions.
The new resolver cases never crossed the config schema, so neither
acceptance path this change adds had an executable gate: config.spec.ts
now refuses an unknown modality at both levels, pins the empty route
list to the namespace validator the settings seam actually runs, and
pins the `[]` materialization the inheritance rule depends on. A
catalog.spec.ts case re-asserts every rung through a written settings
section and `ctx.llm`, which is what the note's Testing section claimed.
Also export PiAiModality from the package entry — it types two public
config fields and consumers could not name it — record the unverified
declaration as a Known Limitation, and stop sharing one fallback array
across a route's models.
A model the installed pi-ai catalog does not describe was reported as
text-only with no way to say otherwise, so a vision model added through
the custom-provider form was refused at every image admission point.
The justification in the source described the DeepSeek chat-completions
serializer, which does reject image blocks; the pi-ai request converter
and every wire protocol it speaks carry images.
Modalities now resolve entry `input` -> installed catalog entry -> route
`defaultInput`, the chain the two capacity fallbacks already use, so the
route value is a fallback and never narrows a catalog model. Its default
stays `[text]`: nothing can interrogate a gateway for its modalities, and
over-claiming admits an image the provider rejects mid-turn, after prompt
admission has already committed the message.
A turn the provider ends at its per-request output cap previously produced
no visible sign in the Web chat flow: only error-kind turn/end events built
a conversation node, so the truncated answer read as a normal completion.
Add a turn-max-tokens conversation node Definition over the durable
turn/end event, a warning-styled localized notice row distinct from
turn-error, a fixture max-tokens sample turn, and an assembled keyless
snapshot pinning the dot state, title, and hint.
Closes#1522
