This PR removes the now-dead direct-routing code:
- Deletes the direct routing implementation.
- Collapses the resolvedModelRoute discriminated union into aiGatewayModelRoute.
- Removes the dead providerKeys cascade.
- Deletes the preferredShortTextCandidates quickgen function.
- Simplifies the advisor override error handling.
- Deprecates the AIGatewayRoutingEnabled deployment option. It is now a no-op so as to not break existing deployments on upgrade.
Once direct routing was gone, the AI Gateway became mandatory for chat, which surfaced gaps in how the product behaves with the gateway disabled:
- Exposes ai-gateway-enabled to the frontend via embedded page metadata.
- Disables the chat composer via the existing AgentSetupNotice when the gateway is disabled, for both new and existing chats.
- Fixes nil/typed-nil chatDaemon panics on startup and shutdown when gateway is disabled.
- Fixes chat WebSocket from retrying the still-gated stream endpoint forever when the gateway is disabled.
Makes the chat context foundation from #26385 live. That PR added the
storage columns, writer queries, and a dormant
`agentapi.ContextDirtyMarker` trigger with no production callers; this
PR wires them together end to end.
When a workspace agent pushes a context snapshot, bound chats now
hydrate to that snapshot's hash, and a later push with a different hash
flips already-pinned chats to dirty (emitting a `context_dirty` watch
event after the transaction commits). Chat creation pins the agent's
latest snapshot when one already exists. The experimental chat API
exposes this as `Chat.Context` (`*ChatContext` with `dirty`,
`dirty_since`, `error`), and a new `PUT
/api/experimental/chats/{chat}/context` endpoint re-pins the agent's
latest snapshot and clears the dirty marker.
`context_dirty_resources` stays NULL (the resource-level diff is
deferred to the UI phase) and the live per-turn context pull is
unchanged.
The end-to-end test provisions a workspace agent via the echo
provisioner, connects it over the Agent API v2.10, and exercises the
full path: an initial push hydrates a bound chat (clean), a second push
with a different hash marks it dirty, the API reports the dirty state,
and the refresh endpoint clears it.
<details>
<summary>Decision log</summary>
- **API shape — sub-struct.** Dirty state is surfaced as
`codersdk.Chat.Context *ChatContext { Dirty bool; DirtySince *time.Time;
Error string }` rather than flat fields, matching the RFC's named
`ChatContext` type and leaving room for future fields (resource diff,
sources). `db2sdk.Chat` populates it when the chat is context-tracked
(`len(ContextAggregateHash) > 0`), dirty, or carries a snapshot error,
and leaves it nil (`omitempty`) otherwise. `Dirty` mirrors
`context_dirty_since` being set.
- **Marker wiring.** The chat daemon is injected directly as the
`agentapi.ContextDirtyMarker`. It is unconditionally constructed (only
its background worker is gated), so the marker is always non-nil and the
wiring matches every other `api.chatDaemon` call site. `agentapi` still
treats a nil marker as "chatd absent", so `PushContextState` stays a
pure write path for any future caller that does not wire chatd in.
- **Refresh is atomic.** `RefreshChatContext` reads the agent's latest
snapshot and re-pins the chat in one repeatable-read transaction, so a
concurrent push cannot land between the read and the write and leave the
chat pinned to a stale hash with the dirty marker cleared.
- **Hydrate + dirty run inside the push transaction.** The fan-out
shares the push's transaction so a concurrent refresh cannot interleave
with the version gate; `context_dirty` watch events publish only after
commit. The pinned hash on dirtied chats is intentionally left unchanged
— the refresh endpoint re-pins it.
- **Dirtied chats keep their pinned hash.** Drift is advisory: a dirty
chat stays usable, and refreshing is the only path that advances the
pinned hash.
- **Test binds `chats.agent_id` directly.** In production the binding is
set lazily during a chat turn (`chatd.persistBuildAgentBinding`); the
test sets it via `dbgen` so it exercises the context flow rather than
turn resolution.
Plan: `coderd/x/chatd` context integration + E2E (sub-struct API,
create-time + push-time hydration, refresh endpoint;
`context_dirty_resources` and the per-turn pull untouched).
</details>
🤖 Generated by Coder Agents on behalf of @kylecarbs
Fixes an issue where sub-agent apps created via CreateSubAgent would
bypass the check for the template's max port sharing level:
- Clamps dynamically inserted `workspace_apps` to the template max
sharing level in `coderd.agentapi.SubAgentAPI`.
- Emits a warning when clamping occurs.
- Adds unit test coverage for the max sharing level matrix.
- Adds an integration-ish test through the devcontainer sub-agent client
path.
> 🤖 Generated by Coder Agents with guidance from a human.
Moves the `coderd_agents_first_connection_seconds` histogram from the
polling-based `prometheusmetrics.Agents()` loop to the event-driven
`agentConnectionMonitor.init()` path. The metric is now recorded exactly
once when an agent first connects over the RPC websocket, instead of
being retroactively computed each polling tick.
The `username` and `workspace_name` labels are removed to reduce
cardinality; only `template_name` and `agent_name` are retained.
Adds unit tests covering both the happy path (first connection recorded)
and the negative-duration guard (clock skew logs a warning, no sample
emitted).
We currently call GetWorkspaceAgentByID millions of times at scale
unnecessarily. This PR embeds immutable fields into the relevant
services instead of fetching for them every time.
resolves https://github.com/coder/scaletest/issues/84
Confirmed with a 10k scaletest that this changeset takes the query from
10M+ queries down to 39k
## Summary
coder-logstream-kube and other tools that use the agent token to connect
to the RPC endpoint were incorrectly triggering connection monitoring,
causing false connected/disconnected timestamps on the agent. This led
to VSCode/JetBrains disconnections and incorrect dashboard status.
## Changes
Add a `role` query parameter to `/api/v2/workspaceagents/me/rpc`:
- `role=agent`: triggers connection monitoring (default for the agent
SDK)
- any other value (e.g. `logstream-kube`): skips connection monitoring
- omitted: triggers monitoring for backward compatibility with older
agents
The agent SDK now sends `role=agent` by default. A new `Role` field on
the `agentsdk.Client` allows non-agent callers to specify a different
role.
## Required follow-up
coder-logstream-kube needs to set `client.Role = "logstream-kube"`
before calling `ConnectRPC20()`. Without that change, it will still send
`role=agent` and trigger monitoring.
Fixes#21625
Adds coderd_template_workspace_build_duration_seconds histogram that
tracks the full duration from workspace build creation to agent ready.
This captures the complete user-perceived build time including
provisioning and agent startup.
The metric is emitted when the agent reports ready/error/timeout via the
lifecycle API, ensuring each build is counted exactly once per replica.
Implements telemetry for boundary usage tracking across all Coder
replicas and reports them via telemetry.
Changes:
- Implement Tracker with Track(), FlushToDB(), and StartFlushLoop() methods
- Add telemetry integration via collectBoundaryUsageSummary()
- Use telemetry lock to ensure only one replica collects per period
The tracker accumulates unique workspaces, unique users, and request
counts (allowed/denied) in memory, then flushes to the database
periodically. During telemetry collection, stats are aggregated across
all replicas and reset for the next period.
Adds template_version_id to re-emitted boundary audit logs to allow
filtering and analysis by specific template versions iin addition to the
existing template_id field. Since boundary policies are defined in the
template, the template version is critical to figuring out which policy
was responsible for boundaries decision in a workspace.
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Fixes all our Go file imports to match the preferred spec that we've _mostly_ been using. For example:
```
import (
"context"
"time"
"github.com/prometheus/client_golang/prometheus"
"golang.org/x/xerrors"
"gopkg.in/natefinch/lumberjack.v2"
"cdr.dev/slog/v3"
"github.com/coder/coder/v2/codersdk/agentsdk"
"github.com/coder/serpent"
)
```
3 groups: standard library, 3rd partly libs, Coder libs.
This PR makes the change across the codebase. The PR in the stack above modifies our formatting to maintain this state of affairs, and is a separate PR so it's possible to review that one in detail.
Upgrades to slog v3 which includes a small, but backward incompatible API change to the acceptible call arguments when logging. This change allows us to verify via compile time type checking that arguments are correct and won't cause a panic, as was possible in slog v1, which this replaces (v2 was tagged but never used in coder/coder).
It also updates dependencies that also use slog and were updated.
I've left the `aibridge` dependency as a commit SHA, under the assumption that the team there (cc @pawbana @dannykopping ) will tag and update the dependency soon and on their own schedule.
Other dependencies, I pushed new tags.
fixes https://github.com/coder/internal/issues/1196
The above issue exposes two different bugs in Coder.
In the agent, there is a race where if the agent is closed while starting up networking, it will erroneously disconnect from Coderd, which delays or breaks writing final status and logs.
In Coderd, there is a bug where we don't properly record the latest agent disconnection time if the agent had previously disconnected. This causes us to report the agent status as "Connected" even after it has disconnected up until the inactivity timeout fires.
This PR fixes both issues.
It also slightly reworks when we send workspace updates based on connection and disconnection. Previously we would send two updates when the agent connected in certain circumstances, even though the status would be the same in both (only times changed). Now we universally only send one on connect, and then another on disconnect.
This PR piggy backs on the agent API cached workspace added in earlier PRs to provide a fast path for avoiding `GetWorkspaceByID` calls in `GetLatestWorkspaceBuildByWorkspaceID` via injection of the workspaces RBAC object into the context. We can do this from the `agentConnectionMonitor` easily since we already cache the workspace.
---------
Signed-off-by: Callum Styan <callumstyan@gmail.com>
The agentapi context needs to be a context with some amount of
authorization attached to it via the context so that the cache refresh
routine can fetch the workspace from the db via GetWorkspaceForAgentID.
---------
Signed-off-by: Callum Styan <callumstyan@gmail.com>
### Breaking Change (changelog note):
> User connections to workspaces, and the opening of workspace apps or ports will no longer create entries in the audit log. Those events will now be included in the 'Connection Log'.
Please see the 'Connection Log' page in the dashboard, and the Connection Log [documentation](https://coder.com/docs/admin/monitoring/connection-logs) for details. Those with permission to view the Audit Log will also be able to view the Connection Log. The new Connection Log has the same licensing restrictions as the Audit Log, and requires a Premium Coder deployment.
### Context
This is the first PR of a few for moving connection events out of the audit log, and into a new database table and web UI page called the 'Connection Log'.
This PR:
- Creates the new table
- Adds and tests queries for inserting and reading, including reading with an RBAC filter.
- Implements the corresponding RBAC changes, such that anyone who can view the audit log can read from the table
- Implements, under the enterprise package, a `ConnectionLogger` abstraction to replace the `Auditor` abstraction for these logs. (No-op'd in AGPL, like the `Auditor`)
- Routes SSH connection and Workspace App events into the new `ConnectionLogger`
- Updates all existing tests to check the values of the `ConnectionLogger` instead of the `Auditor`.
Future PRs:
- Add filtering to the query
- Add an enterprise endpoint to query the new table
- Write a query to delete old events from the audit log, call it from dbpurge.
- Implement a table in the Web UI for viewing connection logs.
> [!NOTE]
> The PRs in this stack obviously won't be (completely) atomic. Whilst they'll each pass CI, the stack is designed to be merged all at once. I'm splitting them up for the sake of those reviewing, and so changes can be reviewed as early as possible. Despite this, it's really hard to make this PR any smaller than it already is. I'll be keeping it in draft until it's actually ready to merge.
Closes https://github.com/coder/internal/issues/619
Implement the `coderd` side of the AgentAPI for the upcoming
dev-container agents work.
`agent/agenttest/client.go` is left unimplemented for a future PR
working to implement the agent side of this feature.
# Use workspace.OwnerUsername instead of fetching the owner
This PR optimizes the agent API by using the `workspace.OwnerUsername` field directly instead of making an additional database query to fetch the owner's username. The change removes the need to call `GetUserByID` in the manifest API and workspace agent RPC endpoints.
An issue arose when the agent token was scoped without access to user data (`api_key_scope = "no_user_data"`), causing the agent to fail to fetch the manifest due to an RBAC issue.
Change-Id: I3b6e7581134e2374b364ee059e3b18ece3d98b41
Signed-off-by: Thomas Kosiewski <tk@coder.com>
This change adds a new `ReportConnection` endpoint to the `agentapi`.
The protocol version was bumped previously, so it has been omitted here.
This allows the agent to report connection events, for example when the
user connects to the workspace via SSH or VS Code.
Updates #15139
Migrates us to `coder/websocket` v1.8.12 rather than `nhooyr/websocket` on an older version.
Works around https://github.com/coder/websocket/issues/504 by adding an explicit test for `xerrors.Is(err, io.EOF)` where we were previously getting `io.EOF` from the netConn.
We currently send empty payloads to pubsub channels of the form `workspace:<workspace_id>` to notify listeners of updates to workspaces (such as for refreshing the workspace dashboard).
To support https://github.com/coder/coder/issues/14716, we'll instead send `WorkspaceEvent` payloads to pubsub channels of the form `workspace_owner:<owner_id>`. This enables a listener to receive events for all workspaces owned by a user.
This PR replaces the usage of the old channels without modifying any existing behaviors.
```
type WorkspaceEvent struct {
Kind WorkspaceEventKind `json:"kind"`
WorkspaceID uuid.UUID `json:"workspace_id" format:"uuid"`
// AgentID is only set for WorkspaceEventKindAgent* events
// (excluding AgentTimeout)
AgentID *uuid.UUID `json:"agent_id,omitempty" format:"uuid"`
}
```
We've defined `WorkspaceEventKind`s based on how the old channel was used, but it's not yet necessary to inspect the types of any of the events, as the existing listeners are designed to fire off any of them.
```
WorkspaceEventKindStateChange WorkspaceEventKind = "state_change"
WorkspaceEventKindStatsUpdate WorkspaceEventKind = "stats_update"
WorkspaceEventKindMetadataUpdate WorkspaceEventKind = "mtd_update"
WorkspaceEventKindAppHealthUpdate WorkspaceEventKind = "app_health"
WorkspaceEventKindAgentLifecycleUpdate WorkspaceEventKind = "agt_lifecycle_update"
WorkspaceEventKindAgentLogsUpdate WorkspaceEventKind = "agt_logs_update"
WorkspaceEventKindAgentConnectionUpdate WorkspaceEventKind = "agt_connection_update"
WorkspaceEventKindAgentLogsOverflow WorkspaceEventKind = "agt_logs_overflow"
WorkspaceEventKindAgentTimeout WorkspaceEventKind = "agt_timeout"
```
A customer hit like 200k of ErrSessionShutdown, which just dupes any errors we would have generated when shutting down the session for e.g. Ping failures.
This adds the ability for `TunnelAuth` to also authorize incoming wireguard node IPs, preventing agents from reporting anything other than their static IP generated from the agent ID.
Moves monitoring of the agent v2 API connection to the yamux layer.
Present behavior monitors this at the websocket layer, and closes the websocket on completion. This can cause yamux to hit unexpected errors since the connection is closed underneath it.
This might be the cause of yamux errors that some customers are seeing
In any case, it's more graceful to close yamux first and let yamux close the underlying websocket. That should limit yamux error logging to truly unexpected/error cases.
The only downside is that the yamux `Close()` doesn't accept a reason, so if the agent becomes outdated and we close the API connection, the agent just sees the connection close without a reason. I'm not sure we log this at the agent anyway, but it would be nice. I think more accurate logging on Coderd are more important.
I've also added some logging when the monitor disconnects for reasons other than the context being canceled (e.g. agent outdated, failed pings).
This PR updates the Agent API to use the appearance.Fetcher, which is set by entitlement code in Enterprise coderd.
This brings the agentapi into compliance with the Enterprise feature.
fixes#10531
Adds a check for `version` on connection to the Agent API websocket endpoint. This is primarily for future-proofing, so that up-level agents get a sensible error if they connect to a back-level Coderd.
It also refactors the location of the `CurrentVersion` variables, to be part of the `proto` packages, since the versions refer to the APIs defined therein.
Refactors the code that handles monitoring an agent websocket with pings and updating the connection times in the DB.
Consolidates v1 and v2 agent APIs under the same code for this.
One substantive change (not _just_ a refactor) is that I've made it so that we actually disconnect if the agent fails to respond to our pings, rather than the old behavior where we would update the database, but not actually tear down the websocket.
Cian Johnston
