fix(custom-model): unset injected env on clear, resume on restart, select the model for pi/omp/grok

Custom Model Endpoint Profiles (#393) let a session point its CLI at a
custom OpenAI-compatible endpoint by injecting env vars or a config file
and restarting the CLI in place. Review of the apply path found four
things, two of them destructive. This lands all four plus the smaller
items from the same review.

1. Clearing a selection did not clear it. The injected vars reach the CLI
   via `tmux setenv`, which persists at the tmux-session level and is
   inherited by `respawn-pane` (measured: `setenv FOO bar` survived two
   successive `respawn-pane -k`), so deleting the keys from the session's
   envOverrides relaunched the CLI still pointed at the old endpoint, and
   for the configDir kinds at a HOME/CODEX_HOME/GROK_HOME that had just
   been deleted. `Session.setCustomModel()` now reports the removed keys,
   queues them (`_pendingEnvUnsets`), and `RespawnPaneOptions.unsetEnvKeys`
   carries them into `applyEnvOverrides()`, which `setenv -u`s them before
   re-applying the live overrides, on the same path that already unsets
   the legacy CLAUDE_CODE_EFFORT_LEVEL. Verified on a private tmux socket
   that `setenv -u HOME` hands the next respawn the global HOME back.

2. Applying a model to a local claude session killed the pane. The
   relaunch was `claude --session-id <id>` and Claude refuses an id that
   already has a transcript, and unlike the dead-pane respawn this one
   kills a working pane first. `restartCli()` now pins the live
   conversation id as the resume id for that respawn when the CLI's launch
   declares a `fallback` chain, which renders the same
   `--resume <id> || --session-id <id>` shape the docker and remote pane
   commands use. Gated on the registry shape, not the CLI id: an entry
   whose resume id is minted by the CLI itself never declares that chain.

3. pi, omp and grok wrote their config file and then launched without the
   `--model` that selects it, so the file was ignored. The registry entry
   now declares `customModelInjection.launchModel` (`custom/{modelId}` for
   pi and omp, grok's `[model.codeman-custom]` block name), the builder
   renders it, and `_withCustomModelLaunchModel()` applies it onto the
   respawn options through `legacyConfigField`, leaving the stored
   <Mode>Config untouched so a clear falls back to the user's own model.
   A model id the CLI's `model` token pattern cannot carry is refused
   with a 400 rather than silently dropped by the argv engine.

4. Remote (SSH) and Docker sessions reported `restarted: true` and changed
   nothing: their `restartCli()` reattaches the durable tmux rather than
   relaunching the agent, and the env lands on the local pane. Both are
   refused with a 400 until those paths are plumbed.

Smaller items from the same review:

- The selection survives a Codeman restart as the disk-only `__customModel`
  bookkeeping (endpoint, model, injected key NAMES, config dir, launch
  model; never the values, which carry the API key). Recovery re-derives
  the values from the endpoint store through the same apply path the route
  uses and keeps the bookkeeping even when the endpoint is gone, so a
  later clear still has keys to unset.
- Discovery goes through `webviewFetch()`, so the RESOLVED address is
  judged by the same egress guard the web-tab proxy uses, and `baseUrl`
  reuses `webviewUrlSchema` (http(s) only, no embedded credentials,
  link-local and cloud-metadata addresses refused). undici's `fetch failed`
  wrapper is unwrapped so the user sees the ECONNREFUSED underneath.
- `custom-model-hosts.json` is written 0600 via tmp+rename, the per-session
  config dir 0700/0600 (pi and omp embed the key literally), and that dir
  is removed with the session.
- `PR.md` is gone from the repo root and the design doc moved to
  `docs/custom-model-endpoints-plan.md` with the LAN address and the
  personal name scrubbed; every reference follows. The guide's `authStyle`
  text matches the shipped schema (`bearer | api-key`, default `bearer`)
  and says that `customModelEndpointsEnabled` is read by nothing until
  the picker lands.
- `config/tsconfig.scripts.json` typechecks `scripts/test-local-llm-harnesses.ts`
  (four real type errors fixed). It is not yet wired into `npm run typecheck`
  because that line differs on master; adding `&& tsc -p config/tsconfig.scripts.json`
  there is the one-line follow-up.

Tests: `test/session-custom-model-restart.test.ts` drives a real Session and
fails on the unfixed code for items 1 to 3; the route suite covers item 4
and the pattern refusal; `test/tmux-manager.test.ts` pins that the unsets
run before the overrides and that a shell-metachar key never reaches tmux.

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This commit is contained in:
Codeman maintainer
2026-09-14 23:46:28 +02:00
parent 1e42cb4e2d
commit 942bf37e48
31 changed files with 988 additions and 457 deletions
+50
View File
@@ -67,6 +67,10 @@ import {
import { imageWatcher } from '../image-watcher.js';
import { workflowRunWatcher, summarizeRun } from '../workflow-run-watcher.js';
import { attachmentRegistry, buildFileThumbnailRoute, registerExternalAttachment } from '../attachment-registry.js';
import { getCli } from '../config/cli-registry/registry.js';
import { readCustomModelHosts } from '../custom-model-hosts.js';
import { applyCustomModelInjection, customModelConfigDir, removeConfigDir } from '../custom-model-injection-apply.js';
import type { CustomModelBookkeeping } from '../types/session.js';
import { registerGeneratedArtifactAttachment } from '../generated-artifact-attachments.js';
import {
buildDetectedAttachmentHistoryItem,
@@ -1181,6 +1185,34 @@ export class WebServer extends EventEmitter {
});
}
/**
* Recovery half of Custom Model Endpoint Profiles: the env values a selection injects
* are never persisted (they carry the API key), so they are computed again from the
* endpoint store, through the SAME apply path the route uses. Undefined when the
* endpoint is gone or the CLI is unregistered: the bookkeeping is still restored so
* the selection can be cleared, and the pane keeps running on tmux's retained env.
*/
private async _rebuildCustomModelEnv(
session: Session,
saved: CustomModelBookkeeping
): Promise<Record<string, string> | undefined> {
const entry = getCli(session.mode);
if (!entry) return undefined;
const endpoint = (await readCustomModelHosts(getDataDir())).find((h) => h.id === saved.endpointId);
if (!endpoint) {
console.warn(
`[WebServer] custom-model endpoint ${saved.endpointId} no longer exists; selection kept for clearing`
);
return undefined;
}
try {
return applyCustomModelInjection(entry, endpoint, saved.modelId, session.id)?.envOverrides;
} catch (err) {
console.warn('[WebServer] Failed to rebuild custom-model env on recovery:', err);
return undefined;
}
}
/** Persists full session state including respawn config to state.json */
private _persistSessionStateNow(session: Session): void {
// See session-manager.updateSessionState: __envOverrides is an internal disk-only
@@ -1190,10 +1222,14 @@ export class WebServer extends EventEmitter {
// __attachmentHistory keeps the private (externalPath-bearing) history on disk,
// separate from the sanitized public attachmentHistory in toState().
const attachmentHistory = session.getAttachmentHistoryForPersist();
// __customModel keeps the selection's bookkeeping (injected env KEYS, config dir,
// launch model; never the values) so recovery can restore and later clear it.
const customModel = session.getCustomModelForPersist();
const state = {
...base,
...(envOverrides ? { __envOverrides: envOverrides } : {}),
...(attachmentHistory ? { __attachmentHistory: attachmentHistory } : {}),
...(customModel ? { __customModel: customModel } : {}),
} as SessionState;
const controller = this.respawnControllers.get(session.id);
if (controller) {
@@ -1396,6 +1432,9 @@ export class WebServer extends EventEmitter {
// come back to a loader whose file we deleted.
if (killMux) {
void removeAgentSessionPreamble(sessionId);
// The per-session custom-model config dir carries the endpoint's API key (pi and
// omp embed it literally); it must not outlive the session it was written for.
removeConfigDir(customModelConfigDir(sessionId));
}
await session.stop(killMux);
this.sessions.delete(sessionId);
@@ -2853,6 +2892,7 @@ export class WebServer extends EventEmitter {
// Note: a legacy CLAUDE_CODE_EFFORT_LEVEL entry is auto-migrated to `effort`
// by the Session constructor (env var would hard-lock /effort switching).
const savedEnvOverrides = (savedState as { __envOverrides?: Record<string, string> })?.__envOverrides;
const savedCustomModel = (savedState as { __customModel?: CustomModelBookkeeping })?.__customModel;
// Prefer the private (externalPath-bearing) history; fall back to the
// sanitized public copy for sessions persisted before that split.
const savedAttachmentHistory =
@@ -2915,6 +2955,16 @@ export class WebServer extends EventEmitter {
parentSessionId: savedState?.parentSessionId,
});
// Custom-model selection survives the restart. The tmux session still carries
// the injected `setenv`s (that is what kept the pane on the endpoint across the
// restart), but `_envOverrides` is rebuilt from a persist that deliberately
// excludes them, so re-derive the values from the endpoint store and re-write
// the isolated config dir; an endpoint that has since been deleted still gets
// the bookkeeping restored, which is what a later clear needs to unset.
if (savedCustomModel) {
session.setCustomModel(savedCustomModel, await this._rebuildCustomModelEnv(session, savedCustomModel));
}
// Update session name if it was a "Restored:" placeholder or doesn't match saved name
if (savedState?.name && muxSession.name !== savedState.name) {
this.mux.updateSessionName(muxSession.sessionId, savedState.name);