[multi-team P3a] Parse and store the team catalog into UserWorkspaces by warp-agent-staging[bot] · Pull Request #15463 · warpdotdev/warp

warp-agent-staging · 2026-08-23T01:47:32Z

Description

Re-split into three, per review. This is PR 1 of a three-PR stack, built directly on master. PR 2 (stacked on this branch) converts the legacy per-user model-list cache into this new storage. PR 3 (stacked on PR 2) moves LLMPreferences off its flat, non-team-aware catalog onto these fields.

Parses and stores the per-team/workspace model catalog into UserWorkspaces. Nothing reads it. LLMPreferences is untouched: it keeps master's flat models_by_feature and behaves exactly as today. This PR is inert by design — dead data plus a conversion test.

Team/Workspace gain a feature_model_choice: ModelsByFeature field.
crates/graphql's get_workspaces_metadata_for_user query gains teams[].featureModelChoice (the workspace-level featureModelChoice already existed), with the matching schema mirror line.
gql_convert::feature_model_choice_from_gql converts the server payload; a malformed catalog is reported and falls back to ModelsByFeature::default() rather than failing the whole Team/Workspace conversion.
New feature_model_choice_json sqlite columns (migration included) persist it through the warm-start cache, so the round trip works before any reader exists.
team_feature_model_choices_conversion_keeps_each_teams_choice_distinct proves each team's uid maps to its own catalog payload, never a shared or swapped one.

Testing

cargo check --workspace --all-targets — clean.
cargo clippy -p warp --lib --tests -- -D warnings — clean.
./script/format — clean (no changes).
./script/bundle --channel oss --nouniversal --check-only — clean.
cargo clippy --locked --target wasm32-unknown-unknown --profile release-wasm-debug_assertions -- -D warnings — clean.
cargo nextest run -p warp -E 'test(gql_convert) or test(user_workspaces) or test(update_manager) or test(team_feature_model_choices)' — 182 passed.
No UI surface changed; this is a data-layer change, so no screenshots.

Linked Issue

N/A — multi-team catalog placement work.

Agent Mode

Warp Agent Mode - This PR was created via Warp's AI Agent Mode

warp-agent-staging · 2026-08-23T01:48:04Z

This PR was generated with Warp.

Comment @warp-factory on this PR to send it follow-up work.

…metadata Rebuilds #15463 on top of master (its old base, #15461, was closed and absorbed into #15446). Partitions LLMPreferences' model catalog by team scope (models_by_team, keyed by team UID with None as the resolved-teamless scope) so every catalog accessor takes a TeamScope instead of reading one ambient, process-wide catalog. Replaces the standalone per-team catalog fetch (which called get_feature_model_choices() with no team and wrote the unscoped result into every team's bucket) with a fold into the existing polled workspaces-metadata query: get_workspaces_metadata_for_user now selects teams[].featureModelChoice for the per-team map, plus the existing workspace-level featureModelChoice for the resolved-teamless case. Both halves arrive on one response, replacing the broken standalone fetch. Trigger disposition: - UserWorkspacesEvent::TeamsChanged no longer triggers a catalog refetch (a workspaces response landing now already carries the catalog); it still runs local custom-model-preference sanitization. - Network-online and AuthComplete no longer trigger a separate catalog fetch: TeamUpdateManager already restarts its poll on network online, and login already kicks off the poller's first tick before AuthComplete fires. - The two ai/blocklist/controller.rs mid-conversation refresh triggers are kept as a narrow, single-scope standalone refresh. Deletes the TODO in llms.rs that asked for exactly this change.

warp-agent-staging · 2026-08-24T22:59:36Z

Converting back to draft — under revision following review (eval-build fix plus four blocking findings). Will mark ready again once addressed and CI is green.

warp-agent-staging · 2026-08-24T23:00:53Z

@IsaiahWitzke one question for you while this is under revision: the review wants visual proof of two teams resolving different catalogs before merge. That needs a staging account whose teams have genuinely different featureModelChoice settings, which I can't provision. Do you want that capture, and can you point me at such an account — or are the two-team regression tests being added here enough evidence for you?

Responding as wilson: Open session · View in factory

…old changes Reduces this PR to PR 1 of a two-PR stack: the model catalog keyed by team (models_by_team, ResolvedTeamScope, the scoped models_by_feature family and its call sites), the fold into the workspaces-metadata query (apply_feature_model_choices, pruning, leave_team), the GraphQL query/schema, the two blocklist refresh triggers rerouted off the unscoped get_feature_model_choices, and the two-team distinctness/ pruning/conversion-level tests. PR 2 (stacked on this branch) adds get_llm_info_for_scope/for_team_uid and threads live TeamScope through the AIExecutionProfileAppExt context-window helpers and RequestParams's request-limit clamp.

Adds Team.feature_model_choice and Workspace.feature_model_choice, sourced from the server's per-team/workspace Team.featureModelChoice (new teams[].featureModelChoice query field, schema mirror line) and the existing Workspace.featureModelChoice, via gql_convert's new feature_model_choice_from_gql. Persisted through new feature_model_choice_json sqlite columns (migration included) so the warm-start cache round-trips it too. Nothing reads this yet -- LLMPreferences keeps master's flat models_by_feature and behaves exactly as today. This PR is inert by design: dead data plus a conversion test (team_feature_model_choices_conversion_keeps_each_teams_choice_distinct) proving each team keeps its own payload through the fold.

… home (#15565) ## Description Stacked on #15463 (PR 1 of the three-PR stack — "Parse and store the team catalog into UserWorkspaces"). PR 1 added `Team`/`Workspace.feature_model_choice` but nothing populates it for a user restored from a pre-upgrade SQLite cache. PR 3 (to follow, stacked on this branch) moves `LLMPreferences` off its flat catalog onto these fields; this PR makes sure a warm-started, pre-upgrade cache has something real to read once that happens. Seeds `Team`/`Workspace.feature_model_choice` from the legacy, pre-team-keyed `MODELS_BY_FEATURE_CACHE_KEY` cache as a one-release migration in `UserWorkspaces::new`: a `Workspace` restored from the SQLite cache predates the `feature_model_choice` column (or the app hasn't fetched since upgrading), so its catalog is still the bare default. This seeds it from the last-known model list instead, so an offline launch right after upgrading isn't stuck on just the `auto` default for the rest of the offline session. `migrate_legacy_feature_model_choices_cache` understands both real shapes an older client could have written: the (more recent) single `ModelsByFeature`, and (older still) a bare `AvailableLLMs`, which becomes the `agent_mode` field. Still nothing reads `Team`/`Workspace.feature_model_choice` for model selection — `LLMPreferences` keeps master's flat `models_by_feature`. `legacy_cache_migration_yields_a_usable_teamless_catalog` proves the older bare-`AvailableLLMs` shape migrates to a catalog whose `agent_mode` bucket resolves the cached model. ## Testing - `cargo check --workspace --all-targets` — clean. - `cargo clippy -p warp --lib --tests -- -D warnings` — clean. - `./script/format` — clean. - `./script/bundle --channel oss --nouniversal --check-only` — clean. - `cargo clippy --locked --target wasm32-unknown-unknown --profile release-wasm-debug_assertions -- -D warnings` — clean. - `cargo nextest run -p warp -E 'test(user_workspaces) or test(legacy_cache) or test(gql_convert)'` — 107 passed. - No UI surface changed; this is a data-layer change, so no screenshots. - This branch is stacked on a non-default base, so it does not get a real CI matrix in this repo; the above was run locally on this exact branch tip. ## Linked Issue - [x] N/A — stacked continuation of #15463. ## Agent Mode - [x] Warp Agent Mode - This PR was created via Warp's AI Agent Mode --------- Co-authored-by: warp-agent-staging[bot] <240773466+warp-agent-staging[bot]@users.noreply.github.com> Co-authored-by: Oz <oz-agent@warp.dev>

…orkspace (#15567) ## Description Stacked on #15565 (PR 2), which is stacked on #15463 (PR 1). This is the third and final PR of the stack: it deletes `LLMPreferences`'s flat, non-team-aware `models_by_feature` field and its `MODELS_BY_FEATURE_CACHE_KEY` read/write, and the old `WorkspacesMetadataResponse.feature_model_choices` field PR 1 deliberately left alone. Every accessor now reads through `UserWorkspaces` instead, resolved against a caller-supplied `TeamScope` (or a raw team uid). - `UserWorkspaces` gains `feature_model_choice_for_scope`/`_for_team_uid` (reads, via the same `scoped_or_workspace_setting` pattern every other team-scoped setting uses; the absent/no-workspace-at-all case falls back to the resolved-teamless `pre_login_models_by_feature` seeded by PR 1/PR 2's migration or a pre-login fetch) and `set_feature_model_choice_for_team_uid` (the narrow one-off authed-refresh write path). - `get_llm_info`/`_for_team_uid`/`_for_scope`, `get_active_base_model`, and the various choice/default getters gained an app/ctx parameter and a `TeamScope` parameter accordingly, with the ~85 call sites across the app updated: `terminal/`, `settings_view/`, `ai/blocklist/`, `ai/agent_sdk/`, `pane_group/`, `search/`, `workspace/`, and `crates/warp_tui/`. - Preserves the earlier review-round behavioral fixes carried since PR 1's first draft: teamless eager fetch for `agent_mode_evals` builds, refresh-through-workspace-metadata (not the unscoped `get_feature_model_choices`) for blocklist triggers, cross-team `get_llm_info` display lookups, and stale-bucket eviction (implied by the wholesale `Team`/`Workspace` replacement on every response). This is the bulk of the three-PR stack by line count, but every design decision here — the storage shape, the legacy-cache conversion — was already reviewed in PR 1 and PR 2; this should read as fan-out from those two, not new design. ### Note on ancestry This branch's only ancestors beyond `master` are PR 1's and PR 2's commits — verified with `git merge-base --is-ancestor` in both directions before pushing, since a three-deep stack is easy to get backwards. ## Testing Since this branch is stacked on a non-default base, it does not get a real CI matrix in this repo (only PR 1 does); all of the below was run locally on this exact branch tip: - `cargo check --workspace --all-targets` — clean. - `cargo clippy -p warp --lib --tests -- -D warnings` — clean. - `./script/format` — clean (2 files needed reformatting, included). - `./script/bundle --channel oss --nouniversal --check-only` — clean. - `cargo clippy --locked --target wasm32-unknown-unknown --profile release-wasm-debug_assertions -- -D warnings` — clean (no dead-code-on-wasm surprises from the call-site fan-out). - `cargo nextest run -p warp -E 'test(llms) or test(execution_profile) or test(update_manager) or test(gql_convert) or test(user_workspaces) or test(profile_model_selector) or test(agent_profiles_page) or test(legacy_cache)'` — 262 passed. - `cargo nextest run --workspace -E 'package(warp_tui) and (test(terminal_session_view) or test(model_menu))'` — 187 passed. - No UI surface changed; this is a data-layer change, so no screenshots. ## Linked Issue - [x] N/A — stacked continuation of #15463 and #15565. ## Agent Mode - [x] Warp Agent Mode - This PR was created via Warp's AI Agent Mode --------- Co-authored-by: warp-agent-staging[bot] <240773466+warp-agent-staging[bot]@users.noreply.github.com> Co-authored-by: Oz <oz-agent@warp.dev> Co-authored-by: Tyler Lam <tyler.lam@warp.dev> Co-authored-by: Warp <agent@warp.dev>

cla-bot Bot added the cla-signed label Aug 23, 2026

warp-agent-staging Bot force-pushed the factory/multi-team-pr3a-catalog-team-scope branch 2 times, most recently from fc2b37d to 84f9212 Compare August 23, 2026 04:32

warp-agent-staging Bot added the factory:wilson label Aug 24, 2026

warp-agent-staging Bot force-pushed the factory/multi-team-pr3a-catalog-team-scope branch from 84f9212 to 49ca77f Compare August 24, 2026 22:29

warp-agent-staging Bot changed the base branch from factory/multi-team-p1b-llms-migration to master August 24, 2026 22:29

warp-agent-staging Bot requested a review from IsaiahWitzke August 24, 2026 22:30

warp-agent-staging Bot changed the title ~~[multi-team P3a] Key the model catalog by team, put a TeamScope on the LLMPreferences singleton boundary~~ [multi-team P3a] Key the model catalog by team, fold into the workspaces-metadata query Aug 24, 2026

warp-agent-staging Bot marked this pull request as ready for review August 24, 2026 22:32

warp-agent-staging Bot marked this pull request as draft August 24, 2026 22:59

warp-agent-staging Bot marked this pull request as ready for review August 24, 2026 23:52

warp-agent-staging Bot mentioned this pull request Aug 26, 2026

[multi-team P3b] Scope LLM lookups and context-window limits to the window's team #15553

Merged

2 tasks

warp-agent-staging Bot force-pushed the factory/multi-team-pr3a-catalog-team-scope branch from f1246dc to 1d8e82a Compare August 26, 2026 04:56

warp-agent-staging Bot mentioned this pull request Aug 26, 2026

[multi-team P3d] Scope LLM lookups and context-window limits to the window's team #15561

Open

2 tasks

warp-agent-staging Bot changed the title ~~[multi-team P3a] Key the model catalog by team, fold into the workspaces-metadata query~~ [multi-team P3a] Move the model catalog from LLMPreferences to Team/Workspace Aug 26, 2026

warp-agent-staging Bot added 2 commits August 26, 2026 05:58

Merge remote-tracking branch 'origin/master' into new-pr1-work

fc3d9d1

warp-agent-staging Bot force-pushed the factory/multi-team-pr3a-catalog-team-scope branch from 1d8e82a to fc3d9d1 Compare August 26, 2026 06:02

warp-agent-staging Bot changed the title ~~[multi-team P3a] Move the model catalog from LLMPreferences to Team/Workspace~~ [multi-team P3a] Parse and store the team catalog into UserWorkspaces Aug 26, 2026

This was referenced Aug 26, 2026

[multi-team P3b] Convert the legacy per-user model cache into the new home #15565

Merged

[multi-team P3c] Move LLMPreferences off its flat catalog onto Team/Workspace #15567

Merged

self review

1c49b5b

IsaiahWitzke approved these changes Aug 26, 2026

View reviewed changes

IsaiahWitzke enabled auto-merge (squash) August 26, 2026 15:53

IsaiahWitzke disabled auto-merge August 26, 2026 16:43

IsaiahWitzke closed this Aug 26, 2026

IsaiahWitzke reopened this Aug 26, 2026

IsaiahWitzke enabled auto-merge (squash) August 26, 2026 16:43

IsaiahWitzke merged commit d6a389d into master Aug 26, 2026
41 of 49 checks passed

IsaiahWitzke deleted the factory/multi-team-pr3a-catalog-team-scope branch August 26, 2026 17:17

Navigation Menu

Uh oh!

[multi-team P3a] Parse and store the team catalog into UserWorkspaces - #15463

[multi-team P3a] Parse and store the team catalog into UserWorkspaces#15463
IsaiahWitzke merged 4 commits into
masterfrom
factory/multi-team-pr3a-catalog-team-scope

warp-agent-staging Bot commented Aug 23, 2026 •

edited

Loading

Uh oh!

warp-agent-staging Bot commented Aug 23, 2026

Uh oh!

warp-agent-staging Bot commented Aug 24, 2026

Uh oh!

warp-agent-staging Bot commented Aug 24, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Uh oh!

Conversation

warp-agent-staging Bot commented Aug 23, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Description

Testing

Linked Issue

Agent Mode

Uh oh!

warp-agent-staging Bot commented Aug 23, 2026

Uh oh!

warp-agent-staging Bot commented Aug 24, 2026

Uh oh!

warp-agent-staging Bot commented Aug 24, 2026

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

warp-agent-staging Bot commented Aug 23, 2026 •

edited

Loading