## Description
Adds Prometheus metrics to the AI Bridge Proxy for observability into
proxy traffic and performance.
## Changes
* Add Metrics struct with the following metrics:
* `connect_sessions_total`: counts CONNECT sessions by type
(mitm/tunneled)
* `mitm_requests_total`: counts MITM requests by provider
* `inflight_mitm_requests`: gauge tracking in-flight requests by
provider
* `mitm_request_duration_seconds`: histogram of request latencies by
provider
* `mitm_responses_total`: counts responses by status code class
(2XX/3XX/4XX/5XX) and provider
* Register metrics with `coder_aibridgeproxyd_` prefix in CLI
* Unregister metrics on server close to prevent registry leaks
* Add `tunneledMiddleware` to track non-allowlisted CONNECT sessions
* Add tests for metric recording in both MITM and tunneled paths
Closes: https://github.com/coder/internal/issues/1185
Adds a new subcommand to print the current session token for use in
scripts and automation, similar to `gh auth token`.
## Usage
```bash
CODER_SESSION_TOKEN=$(coder login token)
```
Fixes#21515
## Description
Add exponential backoff retries to all `go install` and `go mod
download` commands across CI workflows and actions.
## Why
Fixes
[coder/internal#1276](https://github.com/coder/internal/issues/1276) -
CI fails when `sum.golang.org` returns 500 errors during Go module
verification. This is an infrastructure-level flake that can't be
controlled.
## Changes
- Created `.github/scripts/retry.sh` - reusable retry helper with
exponential backoff (2s, 4s, 8s delays, max 3 attempts), using
`scripts/lib.sh` helpers
- Wrapped all `go install` and `go mod download` commands with retry in:
- `.github/actions/setup-go/action.yaml`
- `.github/actions/setup-sqlc/action.yaml`
- `.github/actions/setup-go-tools/action.yaml`
- `.github/workflows/ci.yaml`
- `.github/workflows/release.yaml`
- `.github/workflows/security.yaml`
- Added GNU tools setup (bash 4+, GNU getopt, make 4+) for macOS in
`test-go-pg` job, since `retry.sh` uses `lib.sh` which requires these
tools
## Summary
Fixes the broken Kilo Code documentation link in the AI Bridge
client-config page.
## Changes
- Updated the Kilo Code link from the old
`/docs/features/api-configuration-profiles` (returns 404) to the current
`/docs/ai-providers/openai-compatible` page
The Kilo Code documentation was restructured and the old URL no longer
exists.
Fixes#21750
Fixes: https://github.com/coder/internal/issues/560
"Select" CLI UI component should ignore "space" when `+Add custom value`
is highlighted. Otherwise it interprets that as a potential option...
and panics.
Fixes: coder/internal#767
Adds two new Prometheus metrics for license health monitoring:
- `coderd_license_warnings` - count of active license warnings
- `coderd_license_errors` - count of active license errors
Metrics endpoint after startup of a deployment with license enabled:
```
...
# HELP coderd_license_errors The number of active license errors.
# TYPE coderd_license_errors gauge
coderd_license_errors 0
...
# HELP coderd_license_warnings The number of active license warnings.
# TYPE coderd_license_warnings gauge
coderd_license_warnings 0
...
```
fixes: https://github.com/coder/internal/issues/1304
Subscribe to heartbeats synchronously on startup of PGCoordinator. This ensures tests that send heartbeats don't race with this subscription.
Closes [#1246](https://github.com/coder/internal/issues/1246)
This PR adds a new component to display AI Governance user entitlements
in the Licenses Settings page. The implementation includes:
- New `AIGovernanceUsersConsumptionChart` component that shows the
number of entitled users for AI Governance features
- Storybook stories for various states (default, disabled, error states)
- Integration with the existing license settings page
- Collapsible "Learn more" section with links to relevant documentation
- Updated the ManagedAgentsConsumption component with clearer
terminology ("Agent Workspace Builds" instead of "Managed AI Agents")
The chart displays the number of users entitled to use AI features like
AI Bridge, Boundary, and Tasks, with a note that additional analytics
are coming soon.
### Preview
<img width="3516" height="2390" alt="CleanShot 2026-01-27 at 22 44
25@2x"
src="https://github.com/user-attachments/assets/cb97a215-f054-45cb-a3e7-3055c249ef04"
/>
<img width="3516" height="2390" alt="CleanShot 2026-01-27 at 22 45
04@2x"
src="https://github.com/user-attachments/assets/d2534189-cffb-4ad2-b2e2-67eb045572e8"
/>
---------
Co-authored-by: Jaayden Halko <jaayden.halko@gmail.com>
This pull request makes a minor update to the documentation check
workflow. It clarifies that a comment should not be posted if there are
no documentation changes needed and simplifies the comment format
instructions.
The reaper (PID 1) now returns the child's exit code instead of always
exiting 0. Signal termination uses the standard Unix convention of 128 +
signal number.
fixes#21661
My previous change to this test did not create another **workspace**
using the template containing `coder_ai_task` resources, meaning that
this test was not actually testing the right thing. This PR addresses
this oversight.
The test occasionally times out at 15s on Windows CI runners.
Investigation of CI logs shows the HTTP request to the agent's
gitsshkey endpoint never appears in server logs, suggesting it
hangs before the request completes (possibly in connection setup,
middleware, or database queries). Increase to 60s to reduce flake
rate.
Fixescoder/internal#770
## Description
Moves the provider lookup from `handleRequest` to `authMiddleware` so
that the provider is determined during the `CONNECT` handshake and
stored in the request context. This enables provider information to be
available earlier in the request lifecycle.
## Changes
* Move `aibridgeProviderFromHost` call from `handleRequest` to
`authMiddleware`
* Store `Provider` in `requestContext` during `CONNECT` handshake
* Add provider validation in `authMiddleware` (reject if no provider
mapping)
* Keep defensive provider check in `handleRequest` for safety
Follow-up from: https://github.com/coder/coder/pull/21617
Closes#20598
This pull-request implements a very basic change to also render the
`icon` of the `Preset` when we've specifically defined one within the
template. Furthermore, theres a `ⓘ` icon with a description.
### Preview
<img width="984" height="442" alt="CleanShot 2026-01-27 at 20 15 29@2x"
src="https://github.com/user-attachments/assets/d4ceebf9-a5fe-4df4-a8b2-a8355d6bb25e"
/>
This migration converts all tailnet coordination tables to UNLOGGED:
- `tailnet_coordinators`
- `tailnet_peers`
- `tailnet_tunnels`
UNLOGGED tables skip Write-Ahead Log (WAL) writes, significantly
improving performance for high-frequency updates like coordinator
heartbeats and peer state changes.
The trade-off is that UNLOGGED tables are truncated on crash recovery
and are not replicated to standby servers. This is acceptable for these
tables because the data is ephemeral:
1. Coordinators re-register on startup
2. Peers re-establish connections on reconnect
3. Tunnels are re-created based on current peer state
**Migration notes:**
- Child tables must be converted before the parent table because LOGGED
child tables cannot reference UNLOGGED parent tables (but the reverse is
allowed)
- The down migration reverses the order: parent first, then children
Fixes https://github.com/coder/coder/issues/21333
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.
Closes#21593
Various `<PopoverContent>`'s among the application were found that when
the screen-size was too small we weren't able to actually see the full
content unless we resized the window. This pull-request ensures that the
content is never going to extend past that of the
`--radix-popper-available-height` without having an appropriate
scrollbar.
| Before | After |
| --- | --- |
| <img width="948" height="960" alt="CleanShot 2026-01-21 at 20 56
48@2x"
src="https://github.com/user-attachments/assets/5d15fbf9-1c62-427b-bbed-81239922a6bc"
/> | <img width="896" height="906" alt="CleanShot 2026-01-21 at 21 19
03@2x"
src="https://github.com/user-attachments/assets/cfa5baa5-2ec1-438c-9454-bf3073dc6534"
/> |
Only task workspaces have the checks in wsbuilder for violating the
managed agent caps in the license.
Stopped tasks that are resumed with a regular workspace start **still
count as usage**.
Clarified the definition of Agent Workspace Builds and updated the
previous term used.
<!--
If you have used AI to produce some or all of this PR, please ensure you
have read our [AI Contribution
guidelines](https://coder.com/docs/about/contributing/AI_CONTRIBUTING)
before submitting.
-->
feat: add boundary usage telemetry database schema and RBAC
Adds the foundation for tracking boundary usage telemetry across Coder
replicas. This includes:
- Database schema: `boundary_usage_stats` table with per-replica stats
(unique workspaces, unique users, allowed/denied request counts)
- Database queries: upsert stats, get aggregated summary, reset stats,
delete by replica ID
- RBAC: `boundary_usage` resource type with read/update/delete actions,
accessible only via system `BoundaryUsageTracker` subject (not regular
user roles)
- Tracker skeleton + docs: stub implementation in `coderd/boundaryusage/`
The tracker accumulates stats in memory and periodically flushes to the
database. Stats are aggregated across replicas for telemetry reporting,
then reset when a new reporting period begins. The tracker implementation
and plumbing will be done in a subsequent commit/PR.
---------
Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com>
Changes:
- bump claude-code module version
- add docs for version compatibility (for customers using older version
of coder or claude-code module; before GA)
This undeprecates the `allow-workspace-renames` flag. IIUC, the 'danger'
with using this flag is that the workspace name might have been used in
the definition of some other terraform resources within template code,
so a rename could cause problems such as with persistent disks.
for https://github.com/coder/coder/issues/21628
---------
Signed-off-by: Callum Styan <callumstyan@gmail.com>
Use transition-specific actions when authorizing workspace build
parameter inserts in the database layer so start/stop/delete do not
require workspace.update.
Related to: https://github.com/coder/internal/issues/1299
This pull request updates the `.github/workflows/doc-check.yaml`
workflow to improve how documentation reviews are triggered and handled,
particularly for pull requests that are converted from draft to ready
for review. The changes ensure that documentation checks are performed
at the appropriate times and clarify the workflow's behavior.
**Workflow trigger and logic enhancements:**
* Added support for triggering the documentation check when a pull
request is marked as "ready for review" (converted from draft), both in
the workflow triggers and in the workflow logic.
[[1]](diffhunk://#diff-46e6065a312f35e5d294476e7865089afd10e6072fed80ac77b257e090def149R9)
[[2]](diffhunk://#diff-46e6065a312f35e5d294476e7865089afd10e6072fed80ac77b257e090def149R24)
[[3]](diffhunk://#diff-46e6065a312f35e5d294476e7865089afd10e6072fed80ac77b257e090def149L39-R48)
* Updated the internal context and trigger type handling to recognize
and describe the "ready_for_review" event, providing more accurate
context for the agent.
[[1]](diffhunk://#diff-46e6065a312f35e5d294476e7865089afd10e6072fed80ac77b257e090def149R138-R140)
[[2]](diffhunk://#diff-46e6065a312f35e5d294476e7865089afd10e6072fed80ac77b257e090def149R171-R173)
**Workflow behavior adjustment:**
* Changed the `comment-on-issue` setting to `false`, so the workflow
will no longer automatically comment on the PR issue when running which
was creating unnecessary noise.
Relates to https://github.com/coder/internal/issues/1282
Updates tracking of managed agents to be predicated instead on the
presence of a related `task_id` instead of the presence of a
`coder_ai_task` resource.
## Description
When `CONNECT` requests are missing or have invalid
`Proxy-Authorization` credentials, the proxy now returns a proper `407
Proxy Authentication Required` response with a `Proxy-Authenticate`
challenge header instead of rejecting the connection without an HTTP
response.
Some clients (e.g. Copilot in VS Code) do not send the
`Proxy-Authorization` header on the initial request and rely on
receiving a `407 challenge` to prompt for credentials. Without this fix,
those clients would fail to connect.
## Changes
* Added `newProxyAuthRequiredResponse` helper function to create
consistent `407` responses with the appropriate `Proxy-Authenticate`
header.
* Updated `authMiddleware` to return a `407` challenge instead of
rejecting unauthenticated `CONNECT` requests without an HTTP response
* Refactored `handleRequest` to use the same helper for consistency
* Updated `TestProxy_Authentication` to verify the `407` response
status, `Proxy-Authenticate` header, and response body
Related to: https://github.com/coder/internal/issues/1235
Closes [#1227](https://github.com/coder/internal/issues/1227)
Added support for license addons, starting with AI Governance, to enable
dynamic feature grouping without requiring license reissuance.
### What changed?
- Introduced a new `Addon` type to represent groupings of features that
can be added to licenses
- Created the first addon `AddonAIGovernance` which includes AI Bridge
and Boundary features
- Added validation for addon dependencies to ensure required features
are present
- Added new features: `FeatureBoundary` and
`FeatureAIGovernanceUserLimit`
- Updated license entitlement logic to handle addons and their features
- Added helper methods to check if features belong to addons
- Updated tests to verify addon functionality
### Why make this change?
This change introduces a more flexible licensing model that allows
features to be grouped into addons that can be added to licenses without
requiring reissuance when new features are added to an addon. This is
particularly useful for specialized feature sets like AI Governance,
where related features can be bundled together and sold as a separate
SKU. The addon approach allows for better organization of features and
more granular control over entitlements.
## Summary
Adds documentation for the "disable automatic updates" feature in Coder
Desktop.
This adds a new "Administrator Configuration" section to the Coder
Desktop docs that documents:
- **macOS**: Setting the `disableUpdater` UserDefaults key via MDM or
`defaults` command
- **Windows**: Setting the `Updater:Enable` registry value in
`HKLM\SOFTWARE\Coder Desktop\App`
The feature already exists in both platforms but was not documented in
the user-facing docs.
## Changes
- Added new "Administrator Configuration" section before
"Troubleshooting"
- Documented macOS MDM configuration for disabling updates
- Documented Windows registry configuration for disabling updates
- Mentioned the `ForcedChannel` option for locking update channels
---
Created on behalf of @ethanndickson
---------
Co-authored-by: blink-so[bot] <211532188+blink-so[bot]@users.noreply.github.com>
## Description
Improves logging in `aibridgeproxyd` to provide better observability for
proxy requests. Adds structured logging with request correlation IDs and
propagates request context through the proxy chain.
## Changes
* Add `requestContext` struct to propagate metadata (token, provider,
session ID) through the proxy request/response chain
* ~Add `handleTunnelRequest` to log passthrough requests for
non-allowlisted domains at debug level~ (removed due to verbosity)
* Add `handleResponse` to log responses from `aibridged`
* Log MITM requests routed to `aibridged` at info level, tunneled
requests at debug level
Related to: https://github.com/coder/internal/issues/1185
Closes https://github.com/coder/internal/issues/1167
Previously we were checking that start != end time; this was flaking on
Windows.
On Windows, `time.Now()` has limited resolution (~1ms with Go runtime's
`timeBeginPeriod`, or ~15.6ms in default system resolution). When two
`time.Now()` calls execute within the same clock tick, they return
identical timestamps, causing `StartedAt.Before(EndedAt)` to return
`false`.
**References:**
- [Go issue #8687](https://github.com/golang/go/issues/8687) - Windows
system clock resolution issue
- [Go issue #67066](https://github.com/golang/go/issues/67066) -
time.Now precision on Windows (still open)
Instead, we're changing the assertion to (the more semantically correct)
"end not before start".
A possible future enhancement could be to plumb coder/quartz through the
recording mechanism, but it's unnecessary for now.
Signed-off-by: Danny Kopping <danny@coder.com>
This change adds a POST /workspaceagents/me/tasks/{task}/log-snapshot
endpoint for agents to upload task conversation history during
workspace shutdown. This allows users to view task logs even when the
workspace is stopped.
The endpoint accepts agentapi format payloads (typically last 10
messages, max 64KB), wraps them in a format envelope, and upserts to the
task_snapshots table. Uses agent token auth and validates the task
belongs to the agent's workspace.
Closescoder/internal#1253
The workspace sharing autocomplete was using the site-wide /api/v2/users
endpoint which requires user:read permission. Regular org members don't
have this permission, so they couldn't see other members to share with.
## Sharing Scope
* This iteration of shared workspaces is slated for beta, and the
currently understood use case does not include cross-org workspace
sharing. This can be addressed later if necessary.
## What's Changed
* Changed to use /api/v2/organizations/{org}/members instead, which only
requires organization_member:read permission (already granted to org
members when workspace sharing is enabled).
* Added `OrganizationMemberWithUserData` to the `UserLike` union to
allow for more flexibility in differentiating groups from users
Susana Ferreira