Files
coder/coderd/experiments_test.go
Thomas Kosiewski 9aa7130e01 feat: evaluate experiment rules per user in discovery and chatd (#29964)
Slice 3 of 4 for runtime experiment rules: evaluate stored rules per
user in discovery, the frontend experiments query, and the pilot
`mcp-tool-search` gate in chatd. There is still no way to edit rules
through the product; that comes in slice 4.

With no rules stored and a healthy database, backend decisions match
today. This PR deliberately changes three things:

1. **Fail closed on rule reads.** If the rules read fails,
`mcp-tool-search` is off, even when it is enabled at startup.
2. **Frontend freshness.** The experiments query now refetches.
3. **Caching.** `/api/v2/experiments` and rendered HTML are sent with
`Cache-Control: no-store`. Static assets keep their long-lived caching.

## What's in it

- **coderd** builds `api.ExperimentEvaluator` from
`experiments.NewDBStore` plus the startup `--experiments` list.
- **Discovery:** `GET /api/v2/experiments` and the `<meta
property="experiments">` tag both use `EnabledExperiments(ctx, caller)`.
- **chatd:** `mcp-tool-search` is now user-scoped.
`chatd.Config.ExperimentEvaluator` is required, so `chatd.New` returns
an error without it. `prepareGeneration` decides once per turn for the
chat owner (`chat.OwnerID`); child chats inherit the owner.
- **Ruleguard** (`userScopedExperimentStaticCheck` in
`scripts/rules.go`): rejects static `Experiments.Enabled(...)` and
`httpmw.RequireExperiment*(...)` checks of user-scoped experiments in
server packages. A codersdk test keeps its list equal to
`codersdk.ExperimentsUserScoped`.
- **Frontend:** the experiments query is keyed by user ID and seeded
from the meta tag only when the embedded user matches. It is stale 60 s
after page load and refetches on mount, window focus and reconnect.
Logout also clears the experiments metadata. The dashboard keeps the
last loaded list when a background refetch fails, instead of replacing
the page with an error.
- **Once per turn:** chatd decides `mcp-tool-search` on a turn's first
step and reuses that decision for the turn's later steps, so a rule
change applies from the next turn and never withdraws a `find_tools`
call the model already made. The decision lives on the chat runner,
keyed by the turn's prompt row, like the stop-nudge tracker. A runner
handoff in the middle of a turn evaluates again.
- **Reference audit:** the only non-test server check of
`mcp-tool-search` is in `generation_preparer.go`, and it now goes
through the evaluator.

## Testing

- **Go:** per-user results and `no-store` on `/api/v2/experiments`; the
per-user meta tag and the HTML cache headers in `site`; group and org
conditions end to end in enterprise; chatd owner eligibility, on → off →
on across turns including a child chat, and a failed rules read.
- **Vitest:** user-keyed seeding, staleness from page load after a
remount, refetch on focus, logout clearing, and the dashboard surviving
a failed refetch.

Remote dogfood UAT through Coder Agents passed on a licensed test
deployment:
- per-user discovery in both the API and the meta tag;
- widening a rule, and turning it off, with no restart;
- `off` beating a startup default, and `inherit` restoring it;
- fail closed on malformed JSON and on an unknown mode, with no
condition text in the logs;
- the `no-store` headers;
- real-model chatd enforcement: `find_tools` offered on turn 1, gone
after `off`, back on turn 3.

That round tested `b57207d4ee`. The later commits change only tests, and
a follow-up round will cover the final head.

**For the author:** `coderd/x/chatd/ARCHITECTURE.md`'s list of the
runner's local state doesn't mention the per-turn trackers (debug turn,
session start, stop nudges, and now this experiment decision). I did not
edit that document.

**Known behavior:** a broken rule logs one error line per evaluation
(per request or turn), with only the experiment, revision, category and
position.

## Stack

1. #29869: evaluator library
2. #29898: storage and audit types
3. **This PR:** enforcement and discovery
4. Control plane (experimental API, `coder exp experiment-rules` CLI,
docs)

> [!NOTE]
> This PR was written by an AI agent (Xum).

---
_Generated with [`xum`](https://github.com/coder/xum) • Model:
`anthropic:claude-opus-5-5` • Thinking: `high`_

---------

Signed-off-by: Thomas Kosiewski <tk@coder.com>
2026-09-29 12:23:00 +00:00

191 lines
6.3 KiB
Go

package coderd_test
import (
"context"
"fmt"
"net/http"
"testing"
"github.com/stretchr/testify/require"
"github.com/coder/coder/v2/coderd/coderdtest"
"github.com/coder/coder/v2/coderd/database/dbauthz"
experimentrules "github.com/coder/coder/v2/coderd/experiments"
"github.com/coder/coder/v2/coderd/httpmw"
"github.com/coder/coder/v2/codersdk"
"github.com/coder/coder/v2/testutil"
)
func Test_Experiments(t *testing.T) {
t.Parallel()
t.Run("empty", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
_ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
experiments, err := client.Experiments(ctx)
require.NoError(t, err)
require.NotNil(t, experiments)
require.Empty(t, experiments)
require.False(t, experiments.Enabled("foo"))
})
t.Run("multiple features", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
cfg.Experiments = []string{"foo", "BAR"}
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
_ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
experiments, err := client.Experiments(ctx)
require.NoError(t, err)
require.NotNil(t, experiments)
// Should be lower-cased.
require.ElementsMatch(t, []codersdk.Experiment{"foo", "bar"}, experiments)
require.True(t, experiments.Enabled("foo"))
require.True(t, experiments.Enabled("bar"))
require.False(t, experiments.Enabled("baz"))
})
t.Run("wildcard", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
cfg.Experiments = []string{"*"}
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
_ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
experiments, err := client.Experiments(ctx)
require.NoError(t, err)
require.NotNil(t, experiments)
require.ElementsMatch(t, codersdk.ExperimentsSafe, experiments)
for _, ex := range codersdk.ExperimentsSafe {
require.True(t, experiments.Enabled(ex))
}
require.False(t, experiments.Enabled("danger"))
})
t.Run("alternate wildcard with manual opt-in", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
cfg.Experiments = []string{"*", "dAnGeR"}
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
_ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
experiments, err := client.Experiments(ctx)
require.NoError(t, err)
require.NotNil(t, experiments)
require.ElementsMatch(t, append(codersdk.ExperimentsSafe, "danger"), experiments)
for _, ex := range codersdk.ExperimentsSafe {
require.True(t, experiments.Enabled(ex))
}
require.True(t, experiments.Enabled("danger"))
require.False(t, experiments.Enabled("herebedragons"))
})
t.Run("Unauthorized", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
cfg.Experiments = []string{"*"}
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
// Explicitly omit creating a user so we're unauthorized.
// _ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
_, err := client.Experiments(ctx)
require.Error(t, err)
require.ErrorContains(t, err, httpmw.SignedOutErrorMessage)
})
t.Run("rules per user", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
cfg.Experiments = []string{"foo", string(codersdk.ExperimentMCPToolSearch)}
ownerClient, db := coderdtest.NewWithDatabase(t, &coderdtest.Options{
DeploymentValues: cfg,
})
owner := coderdtest.CreateFirstUser(t, ownerClient)
memberClient, member := coderdtest.CreateAnotherUser(t, ownerClient, owner.OrganizationID)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
revisions := map[codersdk.Experiment]int64{}
writeRule := func(ex codersdk.Experiment, rule experimentrules.Rule) {
t.Helper()
_, stored, _, err := experimentrules.WriteRule(dbauthz.AsSystemRestricted(ctx), db, owner.UserID, ex, rule, revisions[ex])
require.NoError(t, err)
revisions[ex] = stored.Revision
}
requireExperiments := func(client *codersdk.Client, want ...codersdk.Experiment) {
t.Helper()
got, err := client.Experiments(ctx)
require.NoError(t, err)
require.ElementsMatch(t, want, got)
}
// A condition enables a user-scoped experiment only for matching
// users; a kill switch disables a statically enabled one.
writeRule(codersdk.ExperimentExample, experimentrules.Rule{
Mode: experimentrules.ModeCondition,
Condition: fmt.Sprintf("user.username == %q", member.Username),
})
writeRule(codersdk.ExperimentMCPToolSearch, experimentrules.Rule{Mode: experimentrules.ModeOff})
requireExperiments(memberClient, "foo", codersdk.ExperimentExample)
requireExperiments(ownerClient, "foo")
// Inherit restores the startup default.
writeRule(codersdk.ExperimentMCPToolSearch, experimentrules.Rule{Mode: experimentrules.ModeInherit})
requireExperiments(memberClient, "foo", codersdk.ExperimentMCPToolSearch, codersdk.ExperimentExample)
requireExperiments(ownerClient, "foo", codersdk.ExperimentMCPToolSearch)
// The personalized result must not be cached.
res, err := memberClient.Request(ctx, http.MethodGet, "/api/v2/experiments", nil)
require.NoError(t, err)
defer res.Body.Close()
require.Equal(t, http.StatusOK, res.StatusCode)
require.Equal(t, "no-store", res.Header.Get("Cache-Control"))
})
t.Run("available experiments", func(t *testing.T) {
t.Parallel()
cfg := coderdtest.DeploymentValues(t)
client := coderdtest.New(t, &coderdtest.Options{
DeploymentValues: cfg,
})
_ = coderdtest.CreateFirstUser(t, client)
ctx, cancel := context.WithTimeout(context.Background(), testutil.WaitLong)
defer cancel()
experiments, err := client.SafeExperiments(ctx)
require.NoError(t, err)
require.NotNil(t, experiments)
require.ElementsMatch(t, codersdk.ExperimentsSafe, experiments.Safe)
})
}