mirror of
https://github.com/coder/coder.git
synced 2026-11-01 21:13:14 +08:00
d014d4c372fb7dfdb3b0ea00a9935fec9b9528f9
12
Commits
| Author | SHA1 | Message | Date | |
|---|---|---|---|---|
 Michael Suchacz
|
8eb1e43f0b |
feat: apply reasoning effort and surface thinking blocks for Google chat models (#28273)
## Problem
Selecting a reasoning effort on a Google (Gemini) chat model either got
dropped or broke the request outright, and even when the request
succeeded, Gemini chats never showed thinking blocks:
- The native `fantasygoogle` path had no case in `applyReasoningEffort`
at all, so effort was silently ignored. With only `include_thoughts`
configured, Gemini 3+ models default to no thinking, so chats produced
zero reasoning tokens.
- The path chat generation actually uses today routes Google providers
through aibridge as openai-compat clients and forwarded the raw effort
string as `reasoning_effort`. Google's OpenAI-compatible endpoint
validates that value against the model's supported thinking levels
instead of clamping, so out-of-range efforts fail the whole request with
HTTP 400. Verified against the live Gemini API: `gemini-3.1-pro-preview`
accepts low/medium/high and rejects none/minimal/xhigh/max. Because the
config's default effort is sent whenever the user selects nothing, a
config whose default falls outside the model's supported set failed
every no-selection message.
- Google's OpenAI-compatible endpoint never emits thought text for
`reasoning_effort` requests. Thoughts require
`extra_body.google.thinking_config` with `include_thoughts`, which is
mutually exclusive with `reasoning_effort`, and even then the thought
deltas come back inline in `content` (marked
`extra_content.google.thought`, wrapped in `<thought>` markers) rather
than in the `reasoning_content` field OpenAI-compatible clients parse.
So no reasoning parts were ever produced, persisted, or rendered for
Gemini chats (verified live in UAT).
## Fix
Reasoning effort:
- Map the global reasoning effort scale onto Google's `thinking_level`
in `applyReasoningEffort`, gated to Gemini 3+ models and clamped to each
model's supported subset (3.0 Pro launched with LOW/HIGH only, 3.1 Pro
added MEDIUM, Flash accepts all four through 3.6 while 3.7 and the
latest flash aliases drop MINIMAL, image models accept HIGH plus MINIMAL
for flash). Pre-Gemini-3 models reject `thinking_level`, so effort
degrades to a no-op for them and a config-pinned level is dropped at
call time instead of failing every generation.
- Add `thinking_level` to `ChatModelGoogleThinkingConfig` so admins can
pin a level for configs without user-selectable effort. A resolved
per-turn effort overrides the pinned level; an explicit
`thinking_budget` wins over both since the Google API rejects requests
that set `thinking_budget` and `thinking_level` together, and config
validation now rejects that combination up front. The admin panel picks
up the new field automatically through the generated model-options
schema.
Thinking blocks on the openai-compat path (the path chat generation
uses), translated in both directions at the existing Gemini transport
seam:
- chatd's OpenAI-compat request patch swaps `reasoning_effort` for an
equivalent `extra_body.google.thinking_config` with `include_thoughts`
enabled: `thinking_level` on Gemini 3+ (clamped to the model's supported
subset), `thinking_budget` on earlier models (effort `none` maps to
budget 0, which disables thinking). Non-Gemini models and non-Google
routes are untouched.
- The same transport rewrites Google's responses so thought output
surfaces as `reasoning_content`, which fantasy already converts into
reasoning parts: streaming bodies are rewritten per SSE line (preserving
streaming latency), JSON bodies in place, with the
`<thought>`/`</thought>` markers stripped at the transitions.
- aibridge's chat-completions interception preserves the `extra_body`
passthrough that openai-go's typed params silently drop, forwarding it
to Google upstreams only, so the chatd -> aibridge -> Google route
carries the thinking config end to end.
## Validation
- `go test ./coderd/x/chatd/... ./aibridge/... ./internal/googleopenai
./codersdk` pass; golangci-lint clean on touched packages.
- New unit tests cover the effort-to-level mapping for every effort
value, per-model clamping on both paths, the effort-to-thinking_config
request swap (level, budget, pinned-config precedence, non-Gemini
passthrough), SSE and JSON thought-response rewriting (fixtures mirror
live captures, including tool-call and no-thought streams), an
end-to-end fake-SSE test through the fantasy pipeline proving reasoning
deltas emerge with markers stripped, and aibridge `extra_body`
preservation (including proof the typed round trip drops it).
- Verified live against `generativelanguage.googleapis.com` on
`gemini-3-flash-preview` and `gemini-2.5-flash` through the real
chatprovider stack: reasoning deltas stream through with markers
stripped and answer text intact.
- Live UAT on a dogfood deployment (remote Coder Agents chat, 4 rounds):
effort clamping verified end to end (rounds 1-2), the missing-thoughts
limitation isolated (round 3), and the fix validated (round 4):
expandable, readable thinking blocks render for
`gemini-3.1-pro-preview`, `gemini-3-flash-preview`, and
`gemini-2.5-flash` at effort=high; no `<thought>` marker leaks;
effort=none produces no block and no 400; multi-turn history replay
works with a thinking block on each turn.
- `make gen` regenerated `typesGenerated.ts` and
`chatModelOptionsGenerated.json`; `pnpm -C site lint:types` passes.
> 🤖 Mux acted on Mike's behalf to create this pull request.
<!-- mux-attribution: model=claude-opus-4-6 thinking=high -->
|
||
|
Michael Suchacz
|
521c383f6b |
fix: repair stale chat agent bindings after workspace rebuild (#28152)
## Problem When a chat is bound to a workspace, chatd persists `chats.agent_id` pointing at a specific workspace agent, and it only rebinds on the next chat turn. A workspace stop/start creates a new agent with a new ID in the latest build, so the chat page resolves the stale agent ID to `undefined` and the right sidebar silently drops Terminal, Desktop, Browser, apps, and ports even though the workspace is running. The existing read-time enrichment only filled nil agent IDs and skipped stale non-nil ones, so refreshing did not help until the user sent another message. ## Fix - `coderd/exp_chats.go`: single-chat reads now repair agent IDs that no longer resolve in the workspace's latest build, using the same `agentselect.FindChatAgent` selection chatd uses. A repaired binding also carries the latest build's ID so the response never pairs the new agent with the previous build. Bindings that still resolve are preserved, and repair stays best-effort and response-only (no write-on-read). List reads keep the previous nil-fill-only behavior because validating existing bindings would cost a per-workspace authorization lookup per listed chat. - `site/src/pages/AgentsPage/AgentChatPage.tsx`: the workspace watch update handler detects when a running workspace's latest build no longer contains the chat's bound agent and invalidates the chat query once per chat/build/binding key for immediate recovery, and the chat query polls every 30 seconds while the binding remains unresolved so a transiently failed repair retries even when an idle workspace publishes no further watch events. The watch stream replays the current workspace on every (re)connect, so this covers rebuilds that happen while the page is open or disconnected; page loads are covered by the server-side repair. The workspace-watcher bailout now also keys on `latest_build.id` so a rebuild propagates while the page is open. - `site/src/api/queries/chats.ts`: chat watch events replay the persisted (pre-repair) binding, so the summary merge adopts a snapshot's `build_id` only when the snapshot agrees on `agent_id`, keeping the repaired agent/build pair atomic in the caches. ## Testing - `go test ./coderd -run TestEnrichChatAgentIDs` covering repair, keep-valid, selection-error, list-mode-skips-bound, and no-workspaces cases. - Storybook interaction story `RecoversSidebarAfterWorkspaceRebuild` exercising the watch-event to chat-refetch to sidebar-recovery flow (verified red without the invalidation, green with it). - `pnpm test AgentChatPage.test.ts` covering the binding-resolution predicate. > Mux created this PR on Mike's behalf. |
||
|
Michael Suchacz
|
57f38b5c24 |
fix: keep chat attachments while a linking chat exists
Fixes https://linear.app/codercom/issue/CODAGT-616/keep-chat-attachments-while-chats-remain-unarchived Chat attachments could disappear even though the chat was still available. This happened when a message was saved without recording which attachments it used, or when cleanup deleted attachments before an archived chat itself was removed. Creating a chat, sending or queuing a message, and editing a message now record both the message and which attachments it uses as one operation. If the chat is already at the 50-attachment limit, the chat change fails without being partially saved. Concurrent attachment writes serialize the 50-file cap per chat. Cleanup locks candidates and checks again for new links before deleting. If a file becomes unavailable after input validation, create, send, and edit return a clear client error and roll back the chat change. An attachment stays available while any chat that uses it still exists. After an archived chat reaches the end of its retention period and is deleted, an old attachment that no remaining chat uses can be cleaned up. The retention guide and unavailable-attachment UI text document this lifecycle. This change cannot restore attachments that were already deleted. The database migration adds two indexes so attachment cleanup stays fast as attachments accumulate. > This PR was authored by Mux (AI) on Mike's behalf. |
||
|
Michael Suchacz
|
6b8f820493 |
feat: remove native chat cost tracking in favor of AI Gateway cost data (#27330)
## Stack Context This stack makes AI Gateway data and budgets the source of truth for AI spend controls. 1. Re-back the per-chat cost endpoint with AI Gateway data (#27328, merged). 2. Remove native chat usage limits (#27329, merged). 3. **This PR, now based on `main`:** remove native chat cost tracking and its dedicated admin UI. ## Summary Removes native per-message price calculation, model pricing fields, cost persistence, aggregate cost queries, and admin cost API types. It also deletes the Analytics and Spend pages plus their legacy redirects. The AI Gateway-backed per-chat cost row and compact budget indicators remain. The spend documentation is renamed to `spend-management.md` and updated for the remaining surfaces, group budget APIs, CSV export, upgrade handling for native pricing and cost history, and the absence of a deployment-wide spend dashboard. The per-chat cost API documents that data follows AI Gateway retention and reports zero after all matching requests are purged. No schema is dropped in this release. `chat_messages.total_cost_micros` remains nullable and unwritten so replicas from the previous release can continue inserting messages during rolling upgrades. #27600 tracks removal after the compatibility window. > Mux prepared this PR on Mike's behalf. |
||
|
Michael Suchacz
|
c6cee10e8b |
feat: add per-model OpenAI Responses API toggle (#27683)
chatd hardcoded `WithUseResponsesAPI()`, so the provider SDK's static known-model list decided whether an OpenAI model spoke the Responses API or Chat Completions. A model absent from that list silently fell back to Chat Completions until the fantasy fork was patched. This exposes the SDK's `WithResponsesAPIFunc` hook as a per-model setting, `openai_config.use_responses_api`, stored in the existing `chat_model_configs.options` JSONB. Unset keeps the known-model list, `true` forces Responses, `false` forces Chat Completions. There is no migration. It sits in a new construction-time `openai_config` section rather than in `provider_options.openai` because it selects the API when the client is built, while `provider_options` holds per-request parameters. That placement is also load-bearing: a config setting only this field would otherwise materialize an OpenAI request-options struct and turn on provider-side response storage, since `Store` defaults to true there. Three places independently decided the transport and would silently disagree with the client actually built: | Site | Effect when it disagrees | | --- | --- | | `ModelFromConfig` | the transport being overridden | | `AcceptsFilePartMediaType` | text attachments dropped, since Responses natively accepts only images and PDFs | | `UsesResponsesOptions` | the SDK type-asserts the concrete options struct, so every OpenAI option is discarded | They share one predicate here, `chatopenai.UsesResponsesAPI`, with the override threaded to each. The rest of the stack removes that threading by resolving the transport once and carrying it. Compaction overrides and the quickgen debug model built clients without `ConfigOptions`, so they now pass it and pick up both this setting and the existing Anthropic beta headers. The toggle also makes transport-conditional option handling admin-switchable, so two hardening changes ride along. `ServiceTierFromChat` now maps every tier the codersdk enum advertises (`auto`, `default`, `flex`, `scale`, `priority`); it previously returned nil for `default` and `scale`, so flipping a model to Responses silently dropped a configured `service_tier` that the API accepts (fantasy forwards the value unchanged). And a new `TestProviderOptionsTransportParity` pins, per `provider_options.openai` field, which transport honors it, against a table in ARCHITECTURE.md, so a field honored on one transport and silently ignored on the other fails the test unless recorded as intentional. Review rounds also caught two lifecycle gaps around the new field. `isZeroChatModelCallConfig` now inspects `OpenAIConfig`, so a stored options blob whose only setting is this toggle survives into GET/list responses instead of reading as `model_config: null`; `TestIsZeroChatModelCallConfigCoversEveryField` sets each config field in isolation and fails if any field is invisible to the zero check. And the model editor's update path sends an explicit empty `model_config` when an edit clears the last field, since an omitted property preserves the stored options server-side; covered by the `EditClearingLastOptionSendsEmptyConfig` story. Azure keeps following the known-model list, because the Azure provider exposes no equivalent hook. The model editor renders Azure with the OpenAI option schema, so instead of shipping a visible but inert control, the option schema generator gains a `providers` struct tag that it emits as `visible_for_providers`. Gating uses the raw provider type rather than the alias table, so the control appears only for openai-typed providers. No hand-written frontend field: the editor renders it from the generated schema. Closes https://linear.app/codercom/issue/CODAGT-874/add-completionsresponses-api-toggle-in-model-editor > Mux prepared this PR on Mike's behalf. |
||
|
Michael Suchacz
|
95a2c2ba02 |
feat: back the per-chat cost endpoint with AI Gateway data (#27328)
## Stack Context
This stack removes native chat cost tracking and native chat usage
limits, making the AI Gateway the single source of AI spend data and
budget enforcement.
1. **This PR:** re-back the per-chat cost endpoint with AI Gateway data.
2. Remove native chat usage limits end to end, rewiring the sidebar
indicator to gateway spend.
3. Remove native chat cost tracking end to end, deleting the
Analytics/Spend cost UI.
## What?
`GET /api/experimental/chats/{chat}/cost` summed
`chat_messages.total_cost_micros`, which native chat cost tracking
maintained. It now aggregates AI Gateway interception data instead, and
has no native fallback.
- New `GetAIBridgeChatCost` query, authorized through the root chat so
members can read their own chat's cost without gaining access to raw
interception rows.
- Response fields renamed: `priced_message_count` -> `request_count`,
`unpriced_messages_having_usage_count` -> `unpriced_request_count`.
- The chat summary sidebar keys its cost cache by root chat, and hides
the cost row where the AI Gateway is off or unlicensed. The root cost is
invalidated when a chat leaves an active status and when a generated
title lands, since title generation bills its own gateway request.
`GetChatModelUsageCostByChatID` and the rest of native cost tracking are
untouched here; PR 3 removes them.
## Why?
Native cost tracking duplicates what the AI Gateway already records, and
the two disagree. Repointing the endpoint first means the cost UI keeps
working while the native implementation is deleted later in the stack.
Two behaviour changes follow from gateway semantics and are intentional:
- **Requests, not messages.** The gateway records interceptions, so
counts are requests. Title-generation traffic now counts.
- **Whole-tree totals.** The gateway records the *spawning* chat's ID as
the interception session ID, so a subagent's requests are attributed to
its immediate parent, not always the root. Only a whole chat tree can be
summed, so the query resolves the root and aggregates the tree, and
every chat in a tree reports the same total. Native returned per-subtree
totals.
## Attribution and counting semantics
The aggregate groups token usage per interception before counting, so
the reported numbers are per request even though a request records one
usage row per provider response:
- `RequestCount` counts finished `Coder Agents` interceptions in the
tree, including unpriced ones.
- `UnpricedRequestCount` counts requests with at least one usage row the
gateway could not price. It is a subset of `RequestCount`.
- `TotalCostMicros` omits only unpriced usage, so a partially priced
request still contributes its priced portion. The sidebar therefore says
`Excludes unpriced usage from N request(s)` rather than claiming whole
requests were dropped.
A recorded cost of zero is a free request, not an unpriced one. Usage
without an effective group is excluded, matching what never reached
`ai_user_daily_spend`.
## Authorization
Reads go through `ExtractChatParam` plus `ResourceChat`, with no
cost-specific RBAC widening. `TestGetChatCost/MemberCanReadOwnChat`
covers a scoped `agents-access` member reading their own chat's cost,
and `MemberCannotReadOtherUsersChat` still asserts 404 for a non-owner.
Plain members without `agents-access` cannot create or read chats at
all, so they never reach this endpoint.
## Known limitation
AI Gateway data has its own retention period, 60 days by default and
configured independently of chat retention, so spend for requests older
than that is no longer reported. A chat whose gateway records have all
been purged reports zero cost, which is indistinguishable from genuinely
free usage under this contract. The endpoint documents the caveat;
#27330 documents it on the Spend Management page.
In-flight interceptions are excluded, since cost is only known once the
response is recorded. A chat's cost therefore lags the active turn by
one request.
## Rebase note
Rebased onto `main` after #27579 removed the `ai-gateway-cost-control`
experiment. The per-chat cost row is now gated on the `aibridge` feature
alone, matching how #27579 degated the other cost-control surfaces.
> Mux prepared this PR on Mike's behalf.
|
||
|
Jaayden HalkoandCursor
|
6f2011af88 |
feat: add chat summary tab in the right sidebar and per-chat cost endpoint (#26649)
Stacked on #26657 (the persisted whole-chat summary backend). Base branch is `chat-summary-62j9`; review/merge that first. Adds a reusable `ChatSummary` component. The summary text is the persisted whole-chat summary (`chat.summary`) introduced by #26657. It is generated asynchronously and may be `null` until the first summary is produced, in which case the popover renders a muted empty state. Live updates arrive via that PR's `chat_summary_change` watch event, which is already merged into the chat caches. Cost is served by a new per-chat endpoint, `GET /api/experimental/chats/{chat}/cost`, which rolls up assistant-message cost across a chat's root and child (subagent) chats and is authorized like the other `{chat}` routes (read on the chat, 404 otherwise). Visual and interaction coverage lives in `ChatSummary.stories.tsx` and `ChatSummaryPopover.stories.tsx` (including populated-summary, empty-state, and cost-loading cases). --------- Co-authored-by: Cursor <cursoragent@cursor.com> |
||
|
Danielle Maywood
|
0f55c283f1 | fix: use backend-selected chat agent for desktop, git, terminal (#26959) | ||
|
Danny Kopping
|
a1330e3a8c |
refactor: rename Ai* database identifiers to AI* (AIGOV-369) (#26327)
Adds `ai` to sqlc's `gen.go.initialisms` in `coderd/database/sqlc.yaml` so the generated DB code follows Go's initialism convention. Adds the matching `ai` -> `AI` case to the dbgen PascalCase helper (`scripts/dbgen/main.go`) so the corresponding `dbmem` / mock identifiers stay in sync. `make gen` regenerates the rest; hand-written call sites that consume DB-generated identifiers (`enterprise/audit/table.go`, `coderd/database/modelmethods.go`, `enterprise/coderd/aigatewaykeys.go`, `coderd/database/dbauthz/*`, etc.) are updated to match. Scope is deliberately limited to the database layer: - `coderd/rbac/*` (resource and scope generators) is untouched — `ResourceAi*` / `ScopeAi*` constants stay on main's casing. - `codersdk/*` (Go SDK) is untouched — `codersdk.ResourceAi*` / `codersdk.APIKeyScopeAi*` constants stay on main's casing, so external Go SDK consumers see no source-level break. - `Aibridge*` identifiers (one SQL token `aibridge`, not `ai_bridge`) are out of scope. On-the-wire values are unchanged: enum strings, RBAC resource type strings, API key scope strings, and JSON tags all stay the same. The HTTP/JSON surface is unaffected. Refs: [AIGOV-369](https://linear.app/codercom/issue/AIGOV-369/change-ai-references-in-coderddatabasemodelsgo-to-ai) 🤖 Generated with [Coder Agents](https://coder.com) |
||
|
Michael Suchacz
|
c4792cf104 |
fix: show Anthropic Opus 4.7+ thinking (#26026)
## Summary - Updates Coder's pinned `github.com/coder/fantasy` fork to include coder/fantasy#39. - Exposes Anthropic `thinking_display` as a typed chat model provider option with `summarized` and `omitted` values. - Validates configured `thinking_display` values and maps them to `fantasyanthropic.ProviderOptions.ThinkingDisplay`. - Regenerates the API/UI option schemas so the admin model config form gets a generated select field. ## Tests - `go mod tidy` - `make gen` - `go test ./codersdk ./coderd/x/chatd/chatprovider ./coderd -run 'TestChatModelProviderOptions|TestAnthropicThinkingDisplayFromChat|TestProviderOptionsFromChatModelConfig_AnthropicThinkingDisplay|TestMergeMissingProviderOptions_AnthropicThinkingDisplay|TestValidateChatModelProviderOptions_AnthropicThinkingDisplay'` - `go test ./coderd/x/chatd/... ./codersdk` - `go test ./coderd -run 'TestValidateChatModelProviderOptions_AnthropicThinkingDisplay'` - `pnpm --dir site exec -- biome lint --error-on-warnings src/api/chatModelOptionsGenerated.json src/api/typesGenerated.ts` - pre-commit hook, including fmt, lint, and slim build > Mux working on behalf of Mike. |
||
|
Michael Suchacz
|
502c5acca8 |
fix(coderd): preserve gateway model names (#26039)
OpenAI-compatible gateway providers such as OpenRouter require slash-namespaced model IDs to reach the intended upstream model, but native OpenAI routing strips those prefixes. Preserve full model IDs for gateway provider types, reject OpenRouter-like providers configured as native `openai` when a slash model would be stripped, and validate chat model config changes under the provider reference lock while still allowing unrelated edits to existing configs. Split from #26005. > Mux created this PR on behalf of Mike. |
||
|
Ethan
|
1203f625b7 |
feat(coderd): accept parameters in start_workspace tool (#24434)
When the chat `start_workspace` tool triggers an active-version upgrade that introduces new required parameters, the build fails with a parameter validation error. Previously this returned a message telling the user to update from the UI — a dead end for the model. This PR lets the model recover inside the chat by: 1. Accepting an optional `parameters` map on `start_workspace` (same schema as `create_workspace`), forwarded as `RichParameterValues`. 2. Returning structured JSON error responses that preserve validation details and the workspace's `template_id`, so the model can call `read_template` to discover what changed. 3. Replacing the UI-only guidance in `exp_chats.go` with model-actionable retry instructions. The expected model flow on an active-version parameter failure is now: ``` start_workspace → fails (structured error with template_id + validations) read_template → discovers new required parameters start_workspace → retries with parameters map → workspace starts ``` <img width="846" height="511" alt="image" src="https://github.com/user-attachments/assets/d18b6864-5970-4225-8da0-0f2ab134ccb4" /> |