Files
kandev/apps/backend/internal/plugins/host.go
T
João Salavisa fae2303db3 feat(plugins): per-plugin settings pages driven by manifest config_schema (#1761)
* feat(plugins): per-plugin settings pages driven by manifest config_schema

Settings > Plugins rows now open /settings/plugins/:id — a detail page
showing the manifest plus a settings form generated from the plugin
author's config_schema. Secret fields (secret: true / format: password)
are masked on the operator API and kept on masked re-submits; a running
plugin restarts on save and reads its config (cleartext secrets included)
through the new Host GetConfig gRPC RPC.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): write plugin config file owner-only (0600)

Operator config can now carry cleartext secret values (config_schema
fields marked secret), so <id>.config.yml must not be world-readable —
match the 0600 mode the record file already uses.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* feat(plugins): vault-backed plugin secrets — Get/Set/DeleteSecret RPCs and encrypted config secrets

Plugins get vault-backed secret primitives on the Host API
(GetSecret/SetSecret/DeleteSecret, capability-gated, namespaced to
plugin:<id>㊙️<key> so a plugin can only touch its own entries;
purged on uninstall). Secret config_schema fields now live in kandev's
AES-256-GCM vault too — <id>.config.yml persists only a vault reference,
and the Host GetConfig RPC resolves it back to cleartext for the plugin.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* chore(plugins): serve PR screenshots from the media/pr-1761-screenshots ref instead of the repo

The PR body now points at raw URLs pinned to an unmerged media ref, so
the binaries never land in main (PRs are squash-merged; only the branch
tip tree merges).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): address review feedback on settings validation and save flow

- integer form fields reject non-integral input instead of parseInt-truncating
- enum selections keep their schema-declared types (numeric enums round-trip)
- backend enum validation normalizes int-vs-float64 before comparison
- maskSecrets masks any non-zero secret value regardless of runtime type
- save flow splits PATCH from refetch: a refetch failure no longer reports a
  successful save as an error, and typed secrets are masked on screen
- optional enum fields gain a 'Not set' option so they can be cleared
- resolveConfigSecrets copies before mutating; restart-failure SetStatus
  errors are logged; SDK test fake returns a non-nil empty config map
- add missing tests: usePluginConfigForm hook, getPluginConfig client,
  integer/enum regression cases

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): harden secret config invariants from review

- reject non-string values on declared secret fields at validation time,
  closing the path where a numeric/boolean secret bypassed the vault and
  persisted cleartext in the config file
- delete removed secret fields' vault entries only AFTER the config commit
  succeeds, so a failed SetConfig never leaves the still-current config
  referencing a deleted vault entry
- uninstall fails closed: the plugin's vault namespace is purged before any
  destructive step, and a cleanup failure aborts the uninstall (retryable)
  instead of letting a reinstall inherit stale secrets

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): failure-atomic config secrets, quiesced uninstall purge, ctx propagation

- UpdateConfig is now failure-atomic for secret overwrites: storeConfigSecrets
  snapshots each vault entry before writing and returns a rollback that
  restores prior values (or deletes newly-created ones). A vault.Set failure
  mid-loop or a failed SetConfig commit rolls back, so a request reported as
  failed never changes the value a still-current config ref resolves to.
- Uninstall stops the plugin process BEFORE purging its vault namespace, so
  the plugin can no longer race the cleanup with a SetSecret between ListIDs
  and the deletes; the purge stays fail-closed and retryable.
- Thread the request context through Uninstall -> deletePluginSecrets instead
  of context.Background(), per the pass-context-through-chains convention.
- Consolidate the 'secret not found:' prefix check into isSecretNotFound
  (host.go + service rollback share one definition).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): detach rollback context, fail closed without a vault, reconcile uninstall status

- vault rollback closures now run on context.WithoutCancel(ctx): a request
  cancelled mid-save (browser closed) can no longer abort the rollback and
  leave the vault holding new values while the unchanged config still points
  at them (Greptile P1).
- storeConfigSecrets fails closed (errSecretVaultRequired) when a plugin
  declares secret config fields but no vault is wired, instead of silently
  persisting the secret in cleartext.
- vaultRestoreFunc distinguishes not-found from a genuine Reveal error: a
  backend read failure aborts the update before writing rather than risk a
  rollback that deletes a real secret; rollback now returns errors, joined
  into UpdateConfig's failure so an inconsistent vault is surfaced, not just
  logged.
- Uninstall reconciles lifecycle status when a secret-purge failure aborts
  after the process was stopped: a previously-running plugin transitions to
  error and notifies observers instead of being reported active with no
  process running.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): detach removed-secret cleanup ctx and clear stale save dirty flag

- cleanupRemovedConfigSecrets now runs on context.WithoutCancel(ctx), matching
  the rollback path, so a client disconnect right after the config commit can
  no longer cancel the delete and orphan now-unreferenced vault entries.
- tests: assert GetConfig succeeds before treating an empty map as proof that
  nothing persisted (a swallowed read error would be a false pass).
- web: on a post-save refetch failure, rebase initialValues onto the masked
  snapshot so the saved form no longer reads as dirty.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): fail tests on discarded GetConfig errors, drop setState-in-updater

- config_test: capture and fatal on GetConfig errors at the four remaining
  positive-assertion sites (a nil from a failed read would panic on the next
  map access instead of giving a clear failure), matching the earlier fix.
- web hook: compute the masked post-save snapshot from the current values and
  apply it to setValues + setInitialValues directly, instead of calling
  setInitialValues inside a setValues functional updater (a React anti-pattern
  that runs twice under StrictMode).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* fix(plugins): skip validation of a secret field's own vault ref

- validateConfigSchema takes the plugin id and skips a secret field whose
  value is that field's config vault ref (an internal storage marker,
  already validated when first set). Without this, a field declaring both
  secret: true and enum: [...] 400s on every save that keeps the mask,
  because the enum check saw the vault ref instead of a user value. Freshly
  entered secret values are still validated; a foreign plugin's ref is not
  skipped. Covered by unit + service-level regression tests.
- surface mid-loop vault rollback failures in storeConfigSecrets via a
  warning instead of discarding them (double-fault observability).
- web: integer inputs get step=1 so browsers flag non-integral input in
  place (serializeConfigValues remains the backstop).

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

* test(plugins): split config secret tests into config_secrets_test.go

config_test.go had grown past the 800-line backend file-length limit
(revive), failing CI lint. Move the vault-backed config and plugin-scoped
secret tests (plus their fake-vault/store helpers) into a sibling file;
no test logic changes.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>

---------

Co-authored-by: Kandev Agent <agent@kandev.dev>
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
2026-07-17 22:13:58 +01:00

282 lines
9.9 KiB
Go

package plugins
import (
"context"
"encoding/json"
"errors"
"fmt"
"time"
"google.golang.org/grpc/codes"
"google.golang.org/grpc/status"
"github.com/kandev/kandev/internal/events/bus"
"github.com/kandev/kandev/internal/plugins/manifest"
"github.com/kandev/kandev/internal/plugins/state"
"github.com/kandev/kandev/pkg/pluginsdk"
)
// pluginHost implements pluginsdk.Host (the kandev.plugin.v1 Host RPCs, §3
// of docs/plans/plugins/GRPC-CONTRACT.md) for exactly one plugin. Service
// hands runtime.Manager a fresh pluginHost per spawn/restart via
// hostForPlugin, bound to that plugin's id and its manifest-declared
// capabilities at spawn time. Every method is capability-gated: State
// methods (GetState/SetState/DeleteState/ListState) require
// capabilities.state, RevealSecret requires capabilities.secrets. EmitEvent
// is intentionally ungated — the frozen contract's capability-gating list
// (§5) only names state and secrets (api_read/api_write reserved for
// future); event emission has no boolean capability to gate on.
type pluginHost struct {
// UnimplementedHostData is embedded so pluginHost satisfies
// pluginsdk.Host even when one of the data-source fields below is nil
// (e.g. a test pluginHost built without SetDataSources' wiring, or a
// capability the manifest doesn't declare — see host_data.go's denied
// readers for the capability-gated path). Every accessor
// (Tasks/Sessions/Workspaces/Workflows/AgentProfiles/Repositories) is
// overridden with a real, capability-gated implementation in
// host_data.go; this embed only remains as defense-in-depth.
pluginsdk.UnimplementedHostData
pluginID string
capabilities manifest.Capabilities
// configSchema is the plugin's manifest config_schema, used by
// GetConfig to know which fields are secret (and therefore stored as
// vault references to resolve back to cleartext).
configSchema map[string]any
state *state.Store
secrets SecretVault
bus bus.EventBus
// configs reads the plugin's operator-editable config for the ungated
// GetConfig RPC. Satisfied by store.Store; nil in tests that build a
// pluginHost without one (GetConfig then returns an empty map).
configs configReader
// Host data API (ADR 0043) service-layer dependencies, wired by
// Service.hostForPlugin from Service.SetDataSources. See host_data.go.
taskData taskDataSource
workflows workflowLister
workflowSteps workflowStepLister
agentProfiles agentProfileDataSource
sessionCodeStats sessionCodeStatsSource
}
var _ pluginsdk.Host = (*pluginHost)(nil)
// permissionDenied builds the gRPC error RemotePlugin/Host RPCs return for
// an undeclared capability, matching the wire-level message from
// docs/specs/plugins/spec.md ("Permissions"): "capability '<name>' not
// declared".
func permissionDenied(capability string) error {
return status.Errorf(codes.PermissionDenied, "capability '%s' not declared", capability)
}
// taskNotFound builds the gRPC error taskReader.Get returns for an id that
// doesn't resolve to a task — the SAME error, in-process or over the wire
// (grpcHostServer.GetTask forwards it as-is), so a plugin never observes a
// (nil, nil) success for a missing task.
func taskNotFound(id string) error {
return status.Errorf(codes.NotFound, "task %q not found", id)
}
func (h *pluginHost) GetState(ctx context.Context, scope, scopeID, key string) (map[string]any, bool, error) {
if !h.capabilities.State {
return nil, false, permissionDenied("state")
}
raw, found, err := h.state.Get(ctx, h.pluginID, scope, scopeID, key)
if err != nil || !found {
return nil, found, err
}
value, err := unmarshalStateValue(raw)
if err != nil {
return nil, false, err
}
return value, true, nil
}
func (h *pluginHost) SetState(ctx context.Context, scope, scopeID, key string, value map[string]any) error {
if !h.capabilities.State {
return permissionDenied("state")
}
raw, err := json.Marshal(value)
if err != nil {
return fmt.Errorf("plugins: marshal state value: %w", err)
}
return h.state.Set(ctx, h.pluginID, scope, scopeID, key, raw)
}
func (h *pluginHost) DeleteState(ctx context.Context, scope, scopeID, key string) error {
if !h.capabilities.State {
return permissionDenied("state")
}
return h.state.Delete(ctx, h.pluginID, scope, scopeID, key)
}
func (h *pluginHost) ListState(ctx context.Context, scope, scopeID string) ([]pluginsdk.StateEntry, error) {
if !h.capabilities.State {
return nil, permissionDenied("state")
}
entries, err := h.state.List(ctx, h.pluginID, scope, scopeID)
if err != nil {
return nil, err
}
out := make([]pluginsdk.StateEntry, len(entries))
for i, e := range entries {
value, err := unmarshalStateValue(e.Value)
if err != nil {
return nil, err
}
out[i] = pluginsdk.StateEntry{
Key: e.Key,
Value: value,
UpdatedAt: e.UpdatedAt.UTC().Format(time.RFC3339),
}
}
return out, nil
}
// configReader is the narrow slice of store.Store pluginHost needs for the
// GetConfig RPC.
type configReader interface {
GetConfig(id string) (map[string]any, error)
}
// GetConfig returns the plugin's own operator-editable config, secret values
// included — this RPC is how a configured credential (e.g. a PAT) reaches
// the plugin process. Secret fields are stored as vault references
// (configVaultRef) and resolved back to cleartext here; this resolution is
// deliberately NOT gated on capabilities.secrets, since it is the plugin's
// own config, set by the operator specifically for it.
func (h *pluginHost) GetConfig(ctx context.Context) (map[string]any, error) {
if h.configs == nil {
return map[string]any{}, nil
}
config, err := h.configs.GetConfig(h.pluginID)
if err != nil {
return nil, err
}
if config == nil {
return map[string]any{}, nil
}
return h.resolveConfigSecrets(ctx, config)
}
// resolveConfigSecrets replaces each secret config field's vault reference
// with the cleartext value from the vault. A resolution failure is an error
// (not a silent drop): the plugin must never mistake a broken vault for
// "the operator cleared this setting". Works on a copy — writing cleartext
// into the caller's map would leak it into any future cached configReader
// value, even though today's FSStore.GetConfig always returns a fresh map.
func (h *pluginHost) resolveConfigSecrets(ctx context.Context, config map[string]any) (map[string]any, error) {
out := make(map[string]any, len(config))
for k, v := range config {
out[k] = v
}
for field := range secretPropertyKeys(h.configSchema) {
if !isConfigVaultRef(h.pluginID, field, out[field]) {
continue
}
if h.secrets == nil {
return nil, errors.New("plugins: secret vault not configured")
}
cleartext, err := h.secrets.Reveal(ctx, pluginConfigSecretID(h.pluginID, field))
if err != nil {
return nil, fmt.Errorf("plugins: resolve secret config field %q: %w", field, err)
}
out[field] = cleartext
}
return out, nil
}
// GetSecret reads a plugin-owned secret previously stored via SetSecret.
// found is false (with a nil error) when the key was never set. Requires
// capabilities.secrets, like every secret-touching RPC.
func (h *pluginHost) GetSecret(ctx context.Context, key string) (string, bool, error) {
if err := h.checkSecretAccess(key); err != nil {
return "", false, err
}
value, err := h.secrets.Reveal(ctx, pluginSecretID(h.pluginID, key))
if err != nil {
if isSecretNotFound(err) {
return "", false, nil
}
return "", false, err
}
return value, true, nil
}
// SetSecret upserts a plugin-owned secret into kandev's encrypted vault,
// namespaced to this plugin (vault id "plugin:<id>:secret:<key>") — a
// plugin can never write outside its own namespace.
func (h *pluginHost) SetSecret(ctx context.Context, key, value string) error {
if err := h.checkSecretAccess(key); err != nil {
return err
}
vaultID := pluginSecretID(h.pluginID, key)
return h.secrets.Set(ctx, vaultID, vaultID, value)
}
// DeleteSecret removes a plugin-owned secret. Deleting a missing key is not
// an error, matching DeleteState's semantics.
func (h *pluginHost) DeleteSecret(ctx context.Context, key string) error {
if err := h.checkSecretAccess(key); err != nil {
return err
}
err := h.secrets.Delete(ctx, pluginSecretID(h.pluginID, key))
if isSecretNotFound(err) {
return nil
}
return err
}
// checkSecretAccess is the shared gate for the plugin-scoped secret RPCs:
// capability secrets declared, a wired vault, and a key that is a single
// sane identifier (pluginSecretKeyPattern) so it can never smuggle
// separators into the vault-id namespace.
func (h *pluginHost) checkSecretAccess(key string) error {
if !h.capabilities.Secrets {
return permissionDenied("secrets")
}
if h.secrets == nil {
return errors.New("plugins: secret vault not configured")
}
if !pluginSecretKeyPattern.MatchString(key) {
return status.Errorf(codes.InvalidArgument, "invalid secret key %q", key)
}
return nil
}
func (h *pluginHost) RevealSecret(ctx context.Context, ref string) (string, error) {
if !h.capabilities.Secrets {
return "", permissionDenied("secrets")
}
if h.secrets == nil {
return "", errors.New("plugins: secret vault not configured")
}
return h.secrets.Reveal(ctx, ref)
}
// EmitEvent publishes a plugin-originated event onto the bus, subject
// "plugin.<id>.<name>" (per the task's build instructions). A no-op if no
// event bus was wired (e.g. early boot, or a test Service without one).
func (h *pluginHost) EmitEvent(ctx context.Context, name string, payload map[string]any) error {
if h.bus == nil {
return nil
}
subject := "plugin." + h.pluginID + "." + name
event := bus.NewEvent(subject, "plugin:"+h.pluginID, payload)
return h.bus.Publish(ctx, subject, event)
}
// unmarshalStateValue decodes a plugin_state row's JSON value into a
// Go-native map, matching pluginsdk's Struct<->map[string]any convention.
func unmarshalStateValue(raw json.RawMessage) (map[string]any, error) {
var value map[string]any
if err := json.Unmarshal(raw, &value); err != nil {
return nil, fmt.Errorf("plugins: unmarshal state value: %w", err)
}
return value, nil
}