Repository navigation
v0.9.42: organization self-service fixes, Chat admin actions, sandbox timeout output and faster warm calls - #9613
Open
waleedlatif1 wants to merge 27 commits into
Open
waleedlatif1 wants to merge 27 commits into
waleedlatif1 wants to merge 27 commits into
Conversation
waleedlatif1
commented
Nov 3, 2026
waleedlatif1
commented
Collaborator
- fix(db): bootstrap Projects for development push databases that predate them (fix(db): bootstrap Projects for development push databases that predate them #9611)
- fix(organizations): refuse self-leave for directory-managed members and keep the renamed name visible (fix(organizations): refuse self-leave for directory-managed members and keep the renamed name visible #9609)
- improvement(chat): let Chat review organization access requests (improvement(chat): let Chat review organization access requests #9610)
- fix(sim-links): hand off a departing member's links to the successor (fix(sim-links): hand off a departing member's links to the successor #9597)
- improvement(billing): open a Chat plan popup on send instead of a static card above the composer (improvement(billing): open a Chat plan popup on send instead of a static card above the composer #9608)
- fix(knowledge): let members connect their own account to a member-access connector (fix(knowledge): let members connect their own account to a member-access connector #9602)
- fix(organizations): let members leave their organization; remove sign out other devices (fix(organizations): let members leave their organization; remove sign out other devices #9595)
- improvement(workspaces): move workspace updates behind an application operation and let Chat rename workspaces (improvement(workspaces): move workspace updates behind an application operation and let Chat rename workspaces #9607)
- fix(credentials): let members leave a shared credential (fix(credentials): let members leave a shared credential #9605)
- fix(organizations): let admins rename their organization (fix(organizations): let admins rename their organization #9603)
- fix(settings): let people revoke the accounts they connected for an organization, even after leaving (fix(settings): let people revoke the accounts they connected for an organization, even after leaving #9601)
- improvement(chat): let Chat manage sim links, bulk-add permission group members, and preview member removal (improvement(chat): let Chat manage sim links, bulk-add permission group members, and preview member removal #9600)
- improvement(sandbox): verify the workbench CLI inside each command's launch (improvement(sandbox): verify the workbench CLI inside each command's launch #9606)
- fix(ui): disable MCP URL edits and table rename for roles the server refuses (fix(ui): disable MCP URL edits and table rename for roles the server refuses #9599)
- docs(library): update best-self-hosted-ai-workflow-automation-platforms-2026 (docs(library): update best-self-hosted-ai-workflow-automation-platforms-2026 #9598)
- improvement(sandbox): return a timed-out workbench run's output and the files it already wrote (improvement(sandbox): return a timed-out workbench run's output and the files it already wrote #9596)
- docs(library): update best-zapier-alternatives (docs(library): update best-zapier-alternatives #9592)
- improvement(media): record a hosted-key auth failure when fal refuses Sim's account (improvement(media): record a hosted-key auth failure when fal refuses Sim's account #9593)
- docs(library): update best-ai-agent-evaluation-platforms-2026 (docs(library): update best-ai-agent-evaluation-platforms-2026 #9594)
- fix(test): assert recovery backoff on the claims, not worker request arrival (fix(test): assert recovery backoff on the claims, not worker request arrival #9591)
Co-authored-by: Sim Pi Agent <pi@sim.ai>
… Sim's account (#9593) * improvement(media): record a hosted-key auth failure when fal refuses Sim's account Mothership's generate_audio and generate_video call fal on Sim's hosted keys outside the tool executor, so a refused account (revoked key, or a lock for an exhausted balance) only reached logs. The submit path now records a Sim/HostedKey Failed point with reason auth on a 401, 402 or 403, the same series the hosted-key dashboard already plots for executor tools, so an alarm on that reason covers fal too. * improvement(media): label fal auth failures with the refused key and use the shared env-flags mock getFalApiKey now returns the env var its key came from, so a refused key in the numbered pool is named on the metric. The metrics module reads isHosted on each call instead of copying it at import, and the test sets it through the central env-flags mock under fake timers.
* docs(library): update best-zapier-alternatives * Pi Babysit: address PR #9592 feedback --------- Co-authored-by: Sim Pi Agent <pi@sim.ai>
…he files it already wrote (#9596) * improvement(sandbox): return a timed-out workbench run's output and the files it already wrote A run_code call killed at its own timeout returned an empty stdout and only "timed out after Nms", so models reran long jobs from scratch. E2B's SDK throws a stream-less TimeoutError, which dropped the output the stream callbacks had already observed. - E2B adapter: a timed-out command returns the stdout/stderr its callbacks observed - Workbench runs (session sandboxes) that hit their own timeout throw a timeout abort carrying the output tail (64 KB) and the files already in the call's output directory; every existing timeout check still accepts it - The route returns that output in stdout with a 408 and says the workbench and its files persist, so the next call can rerun only the remaining work - Output goes through the same session provenance check and secret-name reporting as any other result; one-shot sandboxes and billing are unchanged * fix(sandbox): keep a cancellation during the timed-out output listing a cancellation
…launch (#9606) Every warm workbench call ran a separate CLI integrity check command (ownership write, launch, stop receipt, settle) before the user's command, about a third of a warm call's fixed cost. - The session supervisor's PID 1 now runs the same check as the ordinary user, after recording itself and before starting the command; a mismatch starts nothing and is reported on the stop receipt the call already fetches - A refused launch is repaired under the session lock and relaunched once; a second refusal fails the call - The check runs with an isolated interpreter instead of the user's login shell, never follows links or blocks on non-regular files, and treats an unreadable artifact as stale so repair overwrites it - A reused machine runs no separate check command; a new machine still installs the CLI first - Gated live test against real E2B covers refusal, repair, hooks, FIFOs, Stop during a slow check, and the warm-call cost
…up members, and preview member removal (#9600)
…rganization, even after leaving (#9601)
* fix(organizations): let admins rename their organization * fix(organizations): keep the rename draft until it is saved or cancelled
* fix(credentials): let members leave a shared credential * fix(credentials): type the stale keys list * fix(credentials): match the last-admin guard to the server
… operation and let Chat rename workspaces (#9607) * improvement(workspaces): move workspace updates behind an application operation and let Chat rename workspaces PATCH/PUT /api/workspaces/[id] now run through defineInternalJsonRoute and the session-only workspaces.update use case, keeping the legacy 403/400/500 bodies, audit changes and response shape. Chat gets a workspace/general settings section backed by workspaces.identity.read and workspaces.identity.update, which admit only name and logo; billing payer and the personal-key policy stay session-only. * fix(workspaces): recheck active state atomically on update The workspace UPDATE now carries the not-archived predicate, so an archival that commits after the use case's reads leaves nothing to write and answers the same 404 as a missing workspace, on both the session route and Chat's identity update. The integration suite holds the row lock, archives the row while the update waits, and asserts the 404 and the unchanged name; the archived-fixture refusal now checks that fixture's own row.
… out other devices (#9595) * fix(organizations): let members leave their organization; remove sign out other devices * fix(organizations): hide Leave when the caller's role is unknown * fix(organizations): offer Leave only from an authoritative role * fix(organizations): use only a freshly fetched summary role for Leave * fix(organizations): refetch the summary role on mount
…ess connector (#9602) * fix(knowledge): let members connect their own account to a member-access connector * fix(knowledge): keep watching member enrollment until it connects
…tic card above the composer (#9608) * improvement(billing): open a Chat plan popup on send instead of a static card above the composer * improvement(billing): name the plan upgrade link input
…9597) * fix(sim-links): hand off a departing member's links to the successor * fix(sim-links): repair leftover links on account deletion * fix(sim-links): lock leftover-link organizations before deletion barriers * fix(sim-links): hold the creator's membership while creating a link
* improvement(chat): let Chat review organization access requests Organization admins can list, preview, approve and deny access requests from an organization conversation through the settings tool's new requests section. The admin list, preview and resolve operations now admit organization-delegated Copilot principals under the sim:settings audience; role, audit, outbox notice and fingerprint checks are unchanged. Request settings stay session-only. * fix(chat): default access request list to pending The settings tool parses list input before calling the reader, so an omitted status reached it as undefined and listed every status. The list schema now defaults status to pending, matching get; explicit statuses still apply.
…nd keep the renamed name visible (#9609) * fix(organizations): refuse self-leave for directory-managed members and keep the renamed name visible * fix(organizations): refuse transfer-and-leave for directory-managed owners * fix(organizations): transfer ownership for directory-managed owners and explain why they stay
…te them (#9611) * fix(db): bootstrap Projects for development push databases that predate them A push-managed database with workspaces but no Project schema could never pass the schema-push guard: push cannot backfill workspace.project_id, and the guard refuses every unenforced Project state. Under SIM_DEV_DB_PUSH=1 the pre-push reconciler now replays migrations 0394 and 0406 (the foundation commits atomically with the rollout marker) and runs the registered 0032_project_membership script migration on its single session, recording it as the migration runner does. An interrupted bootstrap resumes; every other unenforced state still refuses with the same message. The pinned drizzle-kit patch also adds unique constraints before foreign keys, so a composite foreign key on a new unique constraint of an existing table no longer fails, and in the noninteractive push policy it answers the unique-constraint truncate prompt with its non-destructive choice. * fix(db): push the schema before the Project migration reads it The bootstrap ran 0032_project_membership before Drizzle pushed anything, so a development database from before a column the migration reads (such as the fork parent) failed after committing the expansion, on every retry. --prepare now only replays the expansion. push.ts then runs one Drizzle push with workspace.project_id still nullable (SIM_DB_PUSH_PROJECT_BOOTSTRAP=1), runs the migration through reconcile-project-membership.ts --bootstrap, and finishes with the ordinary push. The connector and rollout marker are kept out of Drizzle's diff until the migration retires them. A fresh bootstrap also requires that no public table carries a project_id column. * fix(db): pin the Project bootstrap to the public schema The replayed migration SQL and the script runner name tables unqualified, so a connection search path that lists another schema first created the Project foundation outside public. The refusal snapshot now also covers folders and Projects.
|
The latest updates on your projects. Learn more about Vercel for GitHub. |
![cubic-dev-ai[bot]](https://avatars.githubusercontent.com/in/1082092?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
All reported issues were addressed across 118 files
Reply with feedback, questions, or to request a fix.
Turn on auto-fix | Re-trigger cubic
Contributor
 [High impact] The PR changes organization-wide access workflows and shared sandbox and database paths. The PR appears safe to merge, with a non-blocking gap in partial output for short workbench timeouts. Findings
SummaryThis PR expands organization self-service and Chat administration, and improves persistent workbench calls.
|
![greptile-apps[bot]](https://avatars.githubusercontent.com/in/867647?s=60&v=4){kind=small}
#9615) Chat's settings list advertised the Sim links section to every organization. Its operations already refused when the rollout flag was off, but Chat would still offer it. describeSection now drops it for those organizations, so list omits it and open/get/describe/execute refuse it up front.
…ame the organization's plan on organization routes (#9614)
… settings, scope the Chat plan modal (#9616) - A workbench run with a short budget kept no time for its own timeout to answer, so the call deadline won and the output it printed was lost; hold back a small report reserve and bound the output listing by the time left - Chat listed the Sim links settings section to every organization; a settings adapter can now declare whether its section exists, and Sim links does so from its rollout flag - The Chat plan modal stays with the workspace whose send opened it - Organization conversations ask to upgrade the organization's plan
…ire dev push for the Project bootstrap switch (#9617) - A workbench run with a budget of 20 s or less hands the command its whole budget, so the run's own deadline can fire before the provider's timeout reply. The E2B session adapter now treats that timeout-reason abort like the provider's own timeout and returns the output it already streamed, and throwIfRunTimedOut lets only the run's deadline through, so a Stop stays a cancellation. No budget is taken from short runs. - The nullable workspace.project_id switch now also requires SIM_DEV_DB_PUSH=1, like hnswIndexOptions, so a stray bootstrap flag cannot make an ordinary push or generation drop NOT NULL.
This branch was previously deployed
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.