* ci: move secondary, scheduled and contract workflows to GitHub-hosted runners
Public-repo standard hosted runners are free under Enterprise. Release, build,
benchmark, e2e and unit workflows stay on Blacksmith. ci-xterm-cache seeds both
cache stores since its PR consumer's pool is undecided.
* test(pi): resolve the smoke bundle's external packages from the repo
#34148 pulled zod into the commit-message planner graph; the bundle in the
OS temp dir could not resolve it on any runner.
* ci(unit): use four workers on the hosted four-core x64 runner
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
* ci: move routine checks and unsigned builds to Blacksmith
* ci: retain GitHub Mac images for isolated SSH fixtures
* ci: keep shared terminal perf preparation on Blacksmith
* test: allow encrypted slow-link fixture to finish on ARM CI
---------
Co-authored-by: OrcaWin <293788423+OrcaWin@users.noreply.github.com>
* feat(native-chat): a Pi chat shows the turns made outside it
When a Pi chat's agent opens its conversation, Orca asks Pi for the entries
saved after the point the chat last accounted for (get_entries since that
entry id) and adds the terminal's turns once, in Pi's order, under the
"Continued outside this chat." divider. That point is saved on the
conversation handle after each chat turn and at each open. When the chat's
own unsaved work and the terminal's share a suffix that can't be split
(Orca died mid-turn), that suffix is skipped rather than duplicated, and
later terminal turns show normally. A Pi reply over the 8 MiB line limit now
fails only the optional read that asked for it instead of ending the chat.
* refactor(native-chat): share the catch-up open runner across agents
The Codex catch-up's agent-neutral parts move to one module: the open-time
envelope (its own deadline and cancellation, never gating the open), the
bounded staged batch, the single settlement commit that re-checks the
journal before writing, and the "Continued outside this chat." divider.
Codex keeps only how it reads its saved turns and what the chat already
holds. No behavior change; the other agents' catch-up builds on this
instead of copying it.
* refactor(pi): catch up through the shared catch-up runner instead of a copy
* chore: remove unrelated pnpm executable metadata
* refactor(native-chat): the catch-up runner reads the chat's view only when the agent asks
* refactor(pi): read the chat's view through the runner's view()
* chore: remove regenerated pnpm executable metadata
* fix(native-chat): the catch-up deadline also bounds waiting for its commit to be written
A journal write barrier that never settled after the import was admitted held
the provider's open past the catch-up deadline. The wait now ends with the
deadline or the open's cancellation, like every other step.
* test(pi): check Pi's session entry ids and get_entries since against the pinned real Pi
* test: type the history catch-up sink fixture without assertions
* fix(native-chat): bound and revoke history preparation
* test: the runner's journal fake answers the cursor Pi's view reads
* feat(native-chat): validate expected saved items during history catch-up
* fix(pi): preserve outside history and optional read recovery
* test(pi): fail RPC smoke promptly when the child stops
OrcaWinandOrcaWin
Brennan Benson
Neil