feat(coder): v2.6 Phase 0 — AgentBackend foundations (no behavior change)

Schema, interface, and service scaffold for v2.6 persistent agent sessions.
Nothing in this batch alters runtime behavior.

- schema.sql: add session_worktrees (one shared worktree per session, FK
  sessions(id)) and agent_sessions (one backend session per (session, agent),
  with backend/status CHECKs); add pending_changes.agent column for DiffPanel
  attribution. All three statements idempotent (IF NOT EXISTS).
- services/agent-backend.ts: AgentBackend interface + AgentSessionHandle,
  EnsureSessionOpts, PromptCtx, TurnResult, and the normalized transport-agnostic
  AgentEvent union (text/reasoning/tool_call/tool_update/commands). Types only.
- services/agent-pool.ts: lazy get-or-create AgentPool keyed by
  `${sessionId}:${agent}` + shared `agentPool` singleton. Empty in Phase 0.
- index.ts: widen onClose to await dispatcher.stop() then agentPool.dispose()
  (pool empty, so dispose() is inert).

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

.gitignore

@@ -16,5 +16,5 @@ data/*
 !data/AGENTS.md
 !data/skills/
 !data/mcp.json
-!data/coder-providers.json
+!data/coder-providers.example.json
 codecontext/fork.tar.gz

BOOCODER.md

@@ -44,7 +44,7 @@ BooCoder's coding agents are a **config-backed registry**: built-ins live in `pr
 
 ### Config file: `data/coder-providers.json`
 
-Resolved from `CODER_PROVIDERS_PATH` (default `/data/coder-providers.json`; dev/host path `/opt/boocode/data/coder-providers.json`). It is **tracked in git** via a `.gitignore` exception (the rest of `data/*` is ignored). A missing file, invalid JSON, or a schema mismatch all fall back to built-ins-only — loading never throws at startup.
+Resolved from `CODER_PROVIDERS_PATH` (default `/data/coder-providers.json`; dev/host path `/opt/boocode/data/coder-providers.json`). It is **gitignored** — it's live runtime config that the coder reads *and writes* (UI toggles `PATCH` it), so tracking it would churn `git status`. The tracked reference is `data/coder-providers.example.json`; copy it to `coder-providers.json` to seed overrides. A missing file, invalid JSON, or a schema mismatch all fall back to built-ins-only — loading never throws at startup.
 
 ```json
 {

CHANGELOG.md

@@ -2,6 +2,14 @@
 
 All notable changes per release tag. Most recent on top, ordered by tag creation date (which matches the git history). Tag names follow `vMAJOR.MINOR.PATCH-slug` — the slug describes what shipped, so the tag name alone is enough to recall the batch.
 
+## v2.5.15-acp-path-guard — 2026-05-29
+
+Security fix + repo hygiene. Fixes a path-traversal in the ACP filesystem bridge (`acp-client-fs.ts`, flagged by the automated push security review): the worktree guard used an unbounded `startsWith(resolve(worktreePath))`, so a sibling path sharing the worktree as a string prefix (`<worktree>-evil/…`) escaped the scope — and `writeWorktreeTextFile` writes to disk directly (no `pending_changes` gate), so a confused/buggy ACP agent could write outside its worktree. Now uses a separator-bounded check matching `write_guard.ts` (`resolve()` + `startsWith(root + sep)` / `=== root`) via a shared `resolveInWorktree`, with a regression test covering `../` traversal and the sibling-prefix bug. Symlink-swap/`O_NOFOLLOW` hardening was intentionally skipped — consistent with `write_guard`'s no-realpath stance, and the agent already runs with host FS access so this is a containment guard, not a trust boundary. Separately, stops tracking the live `data/coder-providers.json` (it's runtime config the UI reads *and writes* on provider toggles, which churned `git status`) — it's now gitignored with a tracked `data/coder-providers.example.json` reference; the loader falls back to built-ins-only when the live file is absent. The provider-type duplication (coder ↔ web) stays guarded by the existing text-identity `provider-types-parity.test.ts` — a shared package was considered and declined (drift is already prevented; not worth the Docker/build-order risk at solo scale).
+
+## v2.5.14-claude-md — 2026-05-29
+
+Docs-only — CLAUDE.md session-learnings update, no code. Adds gotchas surfaced while shipping the v2.3 provider-lifecycle batch: the host `boocoder.service` keeps running the old process after `pnpm -C apps/coder build` (stale-process tell = new routes 404 while old routes 200, restart don't re-debug); the `boocode` container `build: .` deploys the working tree, so web edits are live on the Vite dev server but not production until `docker compose up --build -d boocode`; `PATCH /api/providers/config` replaces a provider's override wholesale (send `{...existing, enabled}` or a custom ACP entry's command is wiped) and `data/coder-providers.json` is live config not to be committed as code; external agents dispatch one-shot with no context/token tracking (only native `boocode` tracks ctx; OpenCode-as-server is the unshipped `v2-6-persistent-agent-sessions` plan); the `ui/` primitive inventory with `button role=switch` / Dialog fallbacks for the absent switch/sheet; and the mobile Dialog-with-list scroll-containment recipe. Also backfills previously-uncommitted doc bullets for the `v2.5.7`–`v2.5.11` coder work (provider-type parity test, async ACP command discovery, AgentComposerBar `installed` filter, provider-registry path disambiguation).
+
 ## v2.5.13-provider-lifecycle-phase5 — 2026-05-29
 
 Closeout of the v2.3 provider-lifecycle batch — the web UI (Phase 5) plus docs (Phase 6). Provider management moved into **Settings → Providers**: a tab listing every registered provider with a status badge (Available / Disabled / Not installed / Error / Loading), an enable/disable toggle, a per-provider refresh, and a plaintext diagnostic; toggling sends the provider's *full* override (preserving a custom ACP entry's command under the wholesale-replace PATCH merge) then refetches the snapshot. The composer's provider picker now filters to `enabled && (status === 'ready' || 'loading')`, so disabled and unavailable providers drop out of the picker and are managed only in settings (native `boocode` always shows). A curated ACP catalog (`apps/web/src/data/acp-provider-catalog.ts`) + `AddProviderModal` register custom providers via `PATCH /api/providers/config` then a subset refresh, and the web client gained `getProvidersConfig` / `patchProvidersConfig` / `refreshProviders` / `getProviderDiagnostic`. Two mobile fixes ship alongside: the Settings pane is now reachable on phones (opening it pushes `?pane=` atomically so the mobile URL-sync effect keeps it active instead of snapping back to the chat pane), and the Add-provider modal caps to the viewport with a single `overscroll-contain` scroll region so the list scrolls instead of dragging the whole modal. This completes the arc begun in `v2.5.4-provider-lifecycle-phase1` (config-backed registry over the built-ins) → `v2.5.5-provider-lifecycle-phase2` (loading/unavailable snapshot lifecycle + tier-2 probe TTL gate) → `v2.5.6-provider-lifecycle-phase3` (generic `resolveLaunchSpec` ACP dispatch) → `v2.5.12-provider-lifecycle-phase4` (config GET/PATCH, subset refresh, diagnostic HTTP API). Docs landed in `BOOCODER.md` (config file, refresh contract, enable/disable, custom ACP, the honest subset-refresh known limitation) and `docs/DEFERRED-WORK.md` §2 is marked addressed; the remaining Tier-2 follow-ups (WS `provider_snapshot_updated` frame, `available_agents.enabled` column, shared types package, MCP provider tools) stay deferred.

CLAUDE.md

@@ -68,9 +68,9 @@ Key services:
 - **`messages_with_parts` view** (v1.13.1-B; `schema.sql`). Read sites that need `tool_calls` / `tool_results` / `reasoning_parts` SELECT from this view, NOT `messages` directly. v1.13.20 dropped the legacy `messages.tool_calls` / `messages.tool_results` JSON columns; the view now reads parts-only subselects. Writes target `message_parts` exclusively via `insertParts` (or via the helpers `partsFromAssistantMessage` / `partsFromToolMessage`). The `Message` wire type still carries `tool_calls?` / `tool_results?` because the view synthesizes them from parts — frontend reads are unchanged. Shapes: `tool_calls jsonb[]`, `tool_results jsonb` single object, `reasoning_parts jsonb[]` of `{text}`. If you ever need to UPDATE a message and return its full Message shape, do a two-step UPDATE returning `id` followed by SELECT from the view — RETURNING off the bare `messages` table no longer carries the tool fields.
 - **`services/file_ops.ts`** — Shared file operation implementations used by both inference tools and HTTP routes.
 - **`services/auto_name.ts`** — Non-streaming LLM call to generate 4-word session titles after first assistant reply.
-- **`services/provider-registry.ts`** — Static registry of provider metadata (label, transport, model source). `PROVIDERS` array, `PROVIDERS_BY_NAME` map. 5 providers: boocode (native), opencode (acp), goose (pty), claude (pty), qwen (pty).
-- **`services/agent-probe.ts`** — Startup probe using direct `exec()` (not SSH). Discovers installed agents on host, their versions, ACP support, and models. Qwen models read from `~/.qwen/settings.json`. Claude models are static from the registry. Results persisted to `available_agents` table.
-- **`routes/providers.ts`** — `GET /api/providers` returns installed providers with models. Transport field reflects actual capability (checks `supports_acp` from DB, not just registry preference).
+- **`apps/coder/src/services/provider-registry.ts`** (BooCoder, NOT apps/server) — Static registry of provider metadata (label, transport, model source). `PROVIDERS` array, `PROVIDERS_BY_NAME` map. 5 providers: boocode (native), opencode (acp), goose (pty), claude (pty), qwen (pty).
+- **`apps/coder/src/services/agent-probe.ts`** (BooCoder) — Startup probe using direct `exec()` (not SSH). Discovers installed agents on host, their versions, ACP support, and models. Qwen models read from `~/.qwen/settings.json`. Claude models are static from the registry. Results persisted to `available_agents` table.
+- **`apps/coder/src/routes/providers.ts`** (BooCoder) — `GET /api/providers` returns installed providers with models. Transport field reflects actual capability (checks `supports_acp` from DB, not just registry preference). The apps/server side of this flow is the "Provider picker dispatch" bullet below.
 - **Provider picker dispatch**: when `provider !== 'boocode'`, the message route creates a `tasks` row (with `session_id` set) instead of calling `inference.enqueue`. The dispatcher picks it up and dispatches via ACP or PTY using the agent's `install_path`.
 
 Route registration: all routes registered in `index.ts` via `register*Routes(app, sql, ...)` functions. Routes are in `routes/*.ts`.
@@ -80,12 +80,16 @@ Route registration: all routes registered in `index.ts` via `register*Routes(app
 - Write-capable coding agent. Runs as a **systemd service on the host** (`boocoder.service`), NOT in Docker. Fastify server at port 9502, connects to postgres at `127.0.0.1:5500`.
 - **Workspace dependency on `@boocode/server`**: imports `createInferenceRunner`, `createBroker`, `ALL_TOOLS`, `appendMcpTools` from the server's compiled `dist/`. apps/server's `package.json` has an `exports` map with `types` conditions for NodeNext resolution. apps/server must build FIRST.
 - Build + deploy: `pnpm -C apps/server build && pnpm -C apps/coder build && sudo systemctl restart boocoder`. Env file at `apps/coder/.env.host`. Service file at `/etc/systemd/system/boocoder.service`.
+- After `pnpm -C apps/coder build` the host `boocoder.service` keeps running the OLD process until `sudo systemctl restart boocoder` — a stale process shows **new routes 404 with `{error:'not found'}` while old routes still 200** (the `/api` not-found handler returns that shape). Restart, don't re-debug.
 - Agent dispatch spawns binaries directly using `install_path` from `available_agents` — no `spawn('sh', ['-c', ...])` (fails under systemd). Follows Paseo's pattern: `spawn(fullBinaryPath, argsArray, { cwd })`.
 - systemd hardening: only `NoNewPrivileges=true` is safe. `ProtectSystem`, `ProtectHome`, `PrivateTmp` all break agent dispatch (agents need full filesystem access to read configs, write to worktrees).
 - `apps/server/tsconfig.json` has `declaration: true` so `.d.ts` files exist for workspace consumers.
 - Write tools (`edit_file`, `create_file`, `delete_file`, `apply_pending`, `rewind`) queue in `pending_changes` table. Nothing hits disk until `apply_pending` is called. `write_guard.ts` validates paths (resolve + prefix-check, no realpath since files may not exist for creates).
 - Frontend: NOT a separate SPA. BooCoder is a `'coder'` pane type within BooChat's SPA (`apps/web/`). `CoderPane.tsx` in `apps/web/src/components/panes/`. API requests go through `/api/coder/*` proxy (Vite dev + Fastify production) which rewrites to the boocoder host service (`BOOCODER_URL` env var, default `http://100.114.205.53:9502`). WS connects directly to `:9502`.
 - `apps/coder/web/` is a STANDALONE fallback SPA served at `:9502` directly. The PRIMARY BooCoder frontend is the `CoderPane` in BooChat's SPA (`apps/web/src/components/panes/CoderPane.tsx`), accessible via the "Coder" pane in the workspace at `code.indifferentketchup.com`. Both exist; the pane is what Sam uses.
+- **Provider snapshot lifecycle** (`apps/coder/src/services/`): `provider-config.ts` (Zod config, never-throws on bad input) → `provider-config-registry.ts` (`buildResolvedRegistry`, singleton) → `provider-snapshot.ts` (two-tier probe: tier-1 fast presence, tier-2 cold ACP probe skipped unless force / stale `PROVIDER_PROBE_TTL_MS` 24h / dbEmpty; cached). Verify live: `curl http://100.114.205.53:9502/api/providers/snapshot` — returns providers + models + commands, the exact shape `AgentComposerBar` renders.
+- `PATCH /api/providers/config` replaces a provider id's override object **wholesale** (per-id shallow merge) — to flip one field send `{...existing, enabled}`, or a custom ACP entry's `command`/`label` is wiped and it drops out of the resolved registry. `data/coder-providers.json` is **gitignored** (it's live runtime config — the coder reads AND writes it on UI toggles); the tracked reference is `data/coder-providers.example.json`. The loader falls back to `{providers:{}}` (built-ins only) when the live file is absent, so a fresh checkout needs no copy.
+- External agents dispatch **one-shot** (`opencode acp` / `goose acp` / `qwen --acp`) and report no context-window/token usage; only native `boocode` (llama-swap engine) tracks ctx. OpenCode-as-HTTP-server (warm process + `@opencode-ai/sdk`, the source of a real context bar) is the **planned, unshipped** `openspec/changes/v2-6-persistent-agent-sessions` batch; Paseo's per-provider native clients (design §12) were deliberately not ported.
 
 ### Frontend (`apps/web/src/`)
 
@@ -145,6 +149,7 @@ BooCoder at port 9502: `curl http://100.114.205.53:9502/api/health`. Runs as `bo
 - Tag naming: `vMAJOR.MINOR.PATCH-slug` (e.g. `v1.13.13-ws-publish`). Monotonic per minor — the slug describes the batch's content so the tag name alone is enough to recall what shipped. No letter suffixes (`-a`/`-b`), no pseudo-ranges (`v1.11.x`), no slug-only sub-versions sharing a number (`v1.13.15-tools` + `-openspec` + `-agentlint` — split into sequential patches instead).
 - `CHANGELOG.md` is the per-tag release log, most-recent on top. When a new tag is created, add a `## <tag> — <YYYY-MM-DD>` section with a 3–6 sentence paragraph summarizing what shipped, drawn from the commit body. Cross-reference other tags by name when the batch builds on, fixes, or pairs with prior work (e.g. "pairs with `v1.13.12-ws-schemas`", "fixed in `v1.13.5-stability-bundle`"). No nested bullets — one paragraph.
 - Deploy: `cd /opt/boocode && docker compose up --build -d` (or `docker compose build --no-cache boocode && docker compose up -d` if you suspect a layer-cache issue).
+- The `boocode` container is `build: .` — it builds web+server from the **working tree**, so uncommitted changes deploy. Web edits are live on the Vite dev server (HMR) but NOT on production (`:9500` / code.indifferentketchup.com) until `docker compose up --build -d boocode`.
 - Git push to Gitea: `GIT_SSH_COMMAND="ssh -i /opt/boocode/secrets/boocode_gitea -o IdentitiesOnly=yes" git push origin <branch>`. The default agent identity is rejected; the in-repo deploy key (`secrets/`, gitignored) is the working one. Transient `Connection reset by peer` retries cleanly after `sleep 5`.
 - Don't accumulate `.bak-*` files. Clean them up in the same batch or immediately after merge.
 - DB-integration tests opt-in via env var: `DATABASE_URL='postgres://boocode:devpass@localhost:5500/boochat' pnpm -C apps/server test`. Host port is 5500 (mapped from `boocode_db:5432`); password is `${POSTGRES_PASSWORD}` from `.env` (`devpass`), NOT the literal in `.env`'s `DATABASE_URL=postgres://boocode:Ketchup1479@boocode_db:5432/...` line. `psql` is not on the host PATH — for an interactive query use `docker exec boocode_db psql -U boocode -d boochat -c "..."`. Pattern: `describe.runIf(!!process.env.DATABASE_URL)(...)` with a `beforeAll` that applies the schema via `sql.unsafe(readFileSync(schemaPath))`. Tests skip cleanly when var is unset. `tool_cost_stats.test.ts` is the reference.
@@ -172,10 +177,12 @@ BooCoder at port 9502: `curl http://100.114.205.53:9502/api/health`. Runs as `bo
 - Discriminated unions for type narrowing: `Pane` (by `kind`), `SessionEvent` (by `type`), `InferenceFrame` (by `type`).
 - **Adding a new WS frame type** requires updating BOTH the server's `InferenceFrame` (loose `type:` union + optional fields in `services/inference/turn.ts`) AND the web `WsFrame` (strict discriminated union in `apps/web/src/api/types.ts`). Server publish is permissive; the frontend type is the wire-format gate. The `'usage'` frame added in v1.12.2 needed both sides; missing the web side silently drops the frame at JSON-parse.
 - shadcn primitives live in `components/ui/`. Don't modify them unless adding a new primitive.
+- `ui/` primitives present: button, card, context-menu, dialog, dropdown-menu, input, label, radio-group, sonner, textarea. No switch/sheet/drawer/badge/checkbox — use a `<button role="switch" aria-checked>` toggle (a hand-rolled `Switch` already lives in `SettingsPane.tsx`) and a Dialog-based panel for "drawers".
 - `inferLanguage()` from `lib/attachments.ts` is the canonical file-extension-to-language map. `CodeBlock.tsx` keeps its own `LANG_MAP` because it also resolves markdown fence names.
 - Two UI event buses: `hooks/sessionEvents.ts` for DB-state events (chat_created, session_updated); `lib/events.ts` for ephemeral UI (`sendToTerminal`, `terminalsRegistry`). Don't merge — different subscriber lifecycles.
 - `vite.config.ts` proxy entries are order-sensitive: more-specific prefixes (`/api/term`, `/ws/term`) must come BEFORE `/api`.
 - Mobile pane URL sync (`Session.tsx`): the `?pane=<id>` effect resets `activePaneIdx` whenever `panes` changes. New-pane creation on mobile must push `?pane=` atomically — `addPaneAndSwitch` is the wrapper that does this. `addSplitPane` returns the new pane id for callers.
+- A scrollable list inside a Dialog on mobile: cap `DialogContent` (`max-h-[85vh]` + `grid-rows-[auto_minmax(0,1fr)_auto]`) and make the list the single scroll region with `overscroll-contain` — otherwise touch-scroll drags the whole fixed modal / chains to the page.
 - xterm.js v5 uses canvas rendering — browser doesn't see xterm's selection; the native right-click menu has no working Copy for terminal text. App keybindings (`Cmd/Ctrl-C`, `Cmd/Ctrl-Shift-C`) are the path.
 - **New tools** live in their own `services/<name>.ts` file (see `web_search.ts`, `web_fetch.ts`) — exports a pure `executeFoo(input, ...deps)` for direct test access plus a `ToolDef` wrapper that `loadConfig()`s its real dependencies. Register the ToolDef in `tools.ts` `ALL_TOOLS` (and `READ_ONLY_TOOL_NAMES` if applicable). Inject `fetcher: typeof fetch = fetch` rather than `vi.spyOn(globalThis, 'fetch')` — cleanup is simpler and the production call site stays unchanged.
 - **Sentinels** are `role='system'` rows with structured `metadata.kind` (`cap_hit`, `doom_loop`). UI-only — `buildMessagesPayload` strips them via `isAnySentinel` so the LLM never sees them. A new kind requires arms in `MessageMetadata` in BOTH `apps/server/src/types/api.ts` AND `apps/web/src/api/types.ts`, plus a render branch in `apps/web/src/components/MessageBubble.tsx`.
@@ -191,6 +198,9 @@ BooCoder at port 9502: `curl http://100.114.205.53:9502/api/health`. Runs as `bo
 - **CoderPane uses ChatInput** (`components/panes/CoderPane.tsx`): shares the same `ChatInput` component as BooChat for full parity — attachments, paste-to-chip, auto-grow textarea, queued messages during send. CoderPane's `sendOneMessage` is the send callback; queued messages drain via `useEffect` when `sending` goes false.
 - **Adding a new `SessionEvent` type**: add the interface, add it to the `SessionEvent` union, add a `case` in `useSidebar.ts` `applyEvent` switch (no-op `return prev` is fine), and subscribe in any hook that needs it (e.g. `useSessionStream` for `refetch_messages`).
 - **BooCoder provider registry** (`apps/coder/src/services/provider-registry.ts`): static list of provider defs (boocode, opencode, goose, claude, qwen). `PROBED_AGENT_NAMES` derives from it. Adding/removing providers means editing this file, not the frontend.
+- **AgentComposerBar filters `e.installed`**: provider snapshot entries with `installed:false` (loading/unavailable) are dropped from the dropdown. `getProviderSnapshot` must await the full build — returning synchronous `loading` placeholders makes every provider vanish (the v2.5.7 "no providers showing up" regression); surfacing loading states needs a client poll.
+- **Coder↔web provider-type parity** (`apps/coder/src/services/provider-types.ts` ↔ `apps/web/src/api/types.ts`): enforced by runtime `provider-types-parity.test.ts` (compile-time cross-import is blocked by TS6307 on web's composite tsconfig). Mirror of the ws-frames parity pattern — edit both copies together or the test fails.
+- **ACP command discovery is async**: `acp-probe.ts` must poll after `newSession` for `available_commands_update` (commands arrive in a later notification; reading synchronously captures 0). PTY providers (claude) instead discover from disk via `claude-command-discovery.ts` (`~/.claude/commands` + `enabledPlugins` `skills/`+`commands/`, bare names, deduped). `AgentCommand.kind` tags `'command'` vs `'skill'`; `CoderPane`'s `slashGroups` splits them into icon'd groups. `SlashCommandPicker`'s `groups?` prop is opt-in — BooChat passes flat `items` (unchanged).
 - **Pane header architecture (mobile vs desktop)**: Desktop coder pane header (BooCode label + [+] [×]) lives in `Workspace.tsx` gated by `isCoder && !isMobile`. Mobile coder controls (● ×) live in `Session.tsx` header row next to `MobileTabSwitcher`/`NewPaneMenu`. `AgentComposerBar` (provider/mode/model pickers) renders inside `CoderPane.tsx` on both. The ● status dot is passed via `connected` prop from CoderPane to AgentComposerBar.
 - **MessageBubble shared between BooChat and BooCoder** (`components/MessageBubble.tsx`): accepts optional `actions?: MessageActions` callbacks (onRegenerate, onResend, onFork, onDelete) and `hideActions?: ('fork'|'delete'|'openInPane')[]`. Defaults use BooChat API; CoderPane overrides via `CoderMessageList` props. `CoderTextBubble` was removed. **`CoderMessageList` passes `CoderMessageWire as unknown as Message`** — the coder wire shape lacks `metadata`/`kind`/`summary`, so those fields are `undefined` (not `null`) on coder messages. Null-guards on any `Message` field MUST use loose `!= null`, not strict `!== null` (`undefined !== null` is `true` → `.kind` throws → blank-screen crash). The `as unknown as` cast hides this from tsc; build + typecheck pass while runtime crashes.
 - **llama-sidecar** (`/opt/forks/llama-sidecar/`): Go daemon for per-agent llama-server process pool. Cross-compile: `GOOS=windows GOARCH=amd64 /snap/go/current/bin/go build -o bin/llama-sidecar.exe ./cmd/llama-sidecar`. Gitea: `indifferentketchup/llama-sidecar`. Windows child process gotchas: use `context.Background()` for child lifetime (not request ctx), `os.Open(os.DevNull)` for stdin, `os.Pipe()` for stdout with drain goroutine, `DETACHED_PROCESS | CREATE_NEW_PROCESS_GROUP` creation flags. SSH to sam-desktop: `ssh samki@100.101.41.16`; use `schtasks` for persistent process spawning (SSH `start /B` doesn't survive session close).

apps/coder/src/index.ts

@@ -33,6 +33,7 @@ import { registerProviderRoutes } from './routes/providers.js';
 import { registerWebSocket } from './routes/ws.js';
 // Phase 4: dispatcher + agent probe
 import { createDispatcher } from './services/dispatcher.js';
+import { agentPool } from './services/agent-pool.js';
 import { probeAgents } from './services/agent-probe.js';
 import { getProviderSnapshot, persistProbedModels } from './services/provider-snapshot.js';
 import { setPermissionHooks } from './services/permission-waiter.js';
@@ -178,7 +179,12 @@ async function main() {
   // Phase 4: dispatcher — polls tasks table and runs inference
   const dispatcher = createDispatcher({ sql, inference: inferenceApi, broker, log: app.log, config });
   dispatcher.start();
-  app.addHook('onClose', () => dispatcher.stop());
+  app.addHook('onClose', async () => {
+    // stop() first so in-flight dispatcher turns settle, then drain the pool.
+    // Pool is empty in Phase 0 (nothing spawns yet) — dispose() is inert.
+    await dispatcher.stop();
+    await agentPool.dispose();
+  });
 
   // Register routes
   registerMessageRoutes(app, sql, broker, inferenceApi);

apps/coder/src/schema.sql

@@ -76,6 +76,32 @@ ALTER TABLE tasks ADD COLUMN IF NOT EXISTS mode_id TEXT;
 ALTER TABLE tasks ADD COLUMN IF NOT EXISTS thinking_option_id TEXT;
 ALTER TABLE tasks ADD COLUMN IF NOT EXISTS feature_values JSONB;
 
+-- v2.6: one shared worktree per session (all agents/panes in the session operate in it).
+CREATE TABLE IF NOT EXISTS session_worktrees (
+  session_id UUID PRIMARY KEY REFERENCES sessions(id),
+  worktree_path TEXT NOT NULL,
+  base_commit TEXT,
+  created_at TIMESTAMPTZ NOT NULL DEFAULT clock_timestamp()
+);
+
+-- v2.6: one backend session per (session, agent); resumed on switch-back.
+CREATE TABLE IF NOT EXISTS agent_sessions (
+  session_id UUID NOT NULL REFERENCES sessions(id),
+  agent TEXT NOT NULL,
+  backend TEXT NOT NULL,
+  agent_session_id TEXT,
+  server_port INTEGER,
+  status TEXT NOT NULL DEFAULT 'idle',
+  last_active_at TIMESTAMPTZ,
+  created_at TIMESTAMPTZ NOT NULL DEFAULT clock_timestamp(),
+  PRIMARY KEY (session_id, agent),
+  CONSTRAINT agent_sessions_backend_chk CHECK (backend IN ('opencode_server', 'acp_warm')),
+  CONSTRAINT agent_sessions_status_chk CHECK (status IN ('idle', 'active', 'crashed', 'closed'))
+);
+
+-- v2.6: attribution for DiffPanel badges (Phase 1 UX reads this).
+ALTER TABLE pending_changes ADD COLUMN IF NOT EXISTS agent TEXT;
+
 -- LISTEN/NOTIFY fast path: every tasks INSERT (from any call site — routes,
 -- new_task tool, arena, MCP server) fires pg_notify('tasks_new') in the same
 -- transaction, so the dispatcher reacts immediately instead of waiting for the

apps/coder/src/services/__tests__/acp-client-fs.test.ts

@@ -0,0 +1,50 @@
+import { describe, it, expect, afterEach } from 'vitest';
+import { mkdtempSync, rmSync } from 'node:fs';
+import { tmpdir } from 'node:os';
+import { join } from 'node:path';
+import { readWorktreeTextFile, writeWorktreeTextFile } from '../acp-client-fs.js';
+
+const created: string[] = [];
+function freshWorktree(): string {
+  const wt = mkdtempSync(join(tmpdir(), 'acp-wt-'));
+  created.push(wt);
+  return wt;
+}
+afterEach(() => {
+  for (const d of created.splice(0)) {
+    try {
+      rmSync(d, { recursive: true, force: true });
+      rmSync(`${d}-evil`, { recursive: true, force: true });
+    } catch {
+      /* ignore */
+    }
+  }
+});
+
+describe('acp-client-fs worktree scoping', () => {
+  it('writes then reads a file inside the worktree', async () => {
+    const wt = freshWorktree();
+    await writeWorktreeTextFile(wt, 'sub/dir/note.txt', 'hello');
+    expect(await readWorktreeTextFile(wt, 'sub/dir/note.txt')).toBe('hello');
+  });
+
+  it('rejects ../ traversal on read', async () => {
+    const wt = freshWorktree();
+    await expect(readWorktreeTextFile(wt, '../../etc/passwd')).rejects.toThrow(/escapes worktree/);
+  });
+
+  it('rejects ../ traversal on write', async () => {
+    const wt = freshWorktree();
+    await expect(writeWorktreeTextFile(wt, '../escape.txt', 'x')).rejects.toThrow(/escapes worktree/);
+  });
+
+  it('rejects a sibling-prefix path (the unbounded-startsWith bug)', async () => {
+    const wt = freshWorktree();
+    // Absolute path that shares the worktree as a STRING prefix but is a sibling
+    // dir: `<wt>-evil/...`. A bare `startsWith(<wt>)` wrongly admits it.
+    await expect(readWorktreeTextFile(wt, `${wt}-evil/secret.txt`)).rejects.toThrow(/escapes worktree/);
+    await expect(writeWorktreeTextFile(wt, `${wt}-evil/secret.txt`, 'x')).rejects.toThrow(
+      /escapes worktree/,
+    );
+  });
+});

apps/coder/src/services/acp-client-fs.ts

@@ -1,5 +1,25 @@
 import { promises as fs } from 'node:fs';
-import { dirname, isAbsolute, join, resolve } from 'node:path';
+import { dirname, isAbsolute, resolve, sep } from 'node:path';
+
+/**
+ * Resolve an ACP-supplied path against the agent worktree and reject anything
+ * that escapes it. Mirrors `write_guard.ts`'s check: `resolve()` to normalize
+ * `../` segments, then a **separator-bounded** prefix test — a bare
+ * `startsWith(root)` wrongly admits a sibling dir like `<root>-evil/...`.
+ *
+ * No realpath (consistent with `write_guard.ts`: the target may not exist yet on
+ * write). This is a containment guard for the ACP fs bridge, not a hard trust
+ * boundary — the agent process already runs with host FS access; symlink-swap
+ * hardening (`O_NOFOLLOW`/realpath) is out of scope here.
+ */
+function resolveInWorktree(worktreePath: string, filePath: string): string {
+  const root = resolve(worktreePath);
+  const absolute = isAbsolute(filePath) ? resolve(filePath) : resolve(root, filePath);
+  if (absolute !== root && !absolute.startsWith(root + sep)) {
+    throw new Error(`path escapes worktree: ${filePath}`);
+  }
+  return absolute;
+}
 
 /** Resolve an ACP path against the agent worktree and read a slice of lines. */
 export async function readWorktreeTextFile(
@@ -8,10 +28,7 @@ export async function readWorktreeTextFile(
   line?: number | null,
   limit?: number | null,
 ): Promise<string> {
-  const absolute = isAbsolute(filePath) ? filePath : resolve(worktreePath, filePath);
-  if (!absolute.startsWith(resolve(worktreePath))) {
-    throw new Error(`path escapes worktree: ${filePath}`);
-  }
+  const absolute = resolveInWorktree(worktreePath, filePath);
   const raw = await fs.readFile(absolute, 'utf8');
   if (!line && !limit) return raw;
   const lines = raw.split(/\r?\n/);
@@ -26,10 +43,7 @@ export async function writeWorktreeTextFile(
   filePath: string,
   content: string,
 ): Promise<void> {
-  const absolute = isAbsolute(filePath) ? filePath : resolve(worktreePath, filePath);
-  if (!absolute.startsWith(resolve(worktreePath))) {
-    throw new Error(`path escapes worktree: ${filePath}`);
-  }
+  const absolute = resolveInWorktree(worktreePath, filePath);
   await fs.mkdir(dirname(absolute), { recursive: true });
   await fs.writeFile(absolute, content, 'utf8');
 }

apps/coder/src/services/agent-backend.ts

@@ -0,0 +1,85 @@
+/**
+ * v2.6 — AgentBackend abstraction (Phase 0 scaffold; types only, zero runtime logic).
+ *
+ * The core abstraction for persistent agent sessions. Two implementations land
+ * later: `OpenCodeServerBackend` (Phase 1, opencode HTTP server) and
+ * `WarmAcpBackend` (Phase 2, long-lived ACP process). Backends emit
+ * transport-agnostic `AgentEvent`s; the dispatcher maps them to WS frames.
+ *
+ * Nothing imports this file yet — it must compile standalone.
+ * Spec: openspec/changes/v2-6-persistent-agent-sessions/design.md §2.
+ */
+import type { AcpToolSnapshot } from './acp-tool-snapshot.js';
+import type { AgentCommand } from './provider-types.js';
+
+/** Backend transport kind. Mirrors `agent_sessions.backend` CHECK in schema.sql. */
+export type AgentBackendKind = 'opencode_server' | 'acp_warm';
+
+/**
+ * Normalized, transport-agnostic events a backend emits during a turn (§2).
+ * Derived from acp-dispatch's session-update handling, but WITHOUT the WS
+ * envelope (message_id/chat_id) — the dispatcher owns frame mapping.
+ *
+ * `tool_call` vs `tool_update` are kept distinct on purpose: acp-dispatch
+ * currently merges both into one snapshot frame, but opencode's SSE
+ * distinguishes tool-start from tool-result, so the contract carries both.
+ * `commands` mirrors the ACP `available_commands_update` path (v2.5.10).
+ */
+export type AgentEvent =
+  | { type: 'text'; text: string }
+  | { type: 'reasoning'; text: string }
+  | { type: 'tool_call'; toolCall: AcpToolSnapshot }
+  | { type: 'tool_update'; toolCall: AcpToolSnapshot }
+  | { type: 'commands'; commands: AgentCommand[] };
+
+/** Params to establish (or look up) a backend session (§2). */
+export interface EnsureSessionOpts {
+  agent: string;
+  /** Resolved model id. */
+  model: string;
+  /** Shared per-session worktree (one per `sessions.id`, not per pane). */
+  worktreePath: string;
+  projectId: string;
+}
+
+/** Opaque handle to a live backend session, persisted to `agent_sessions` (§2). */
+export interface AgentSessionHandle {
+  sessionId: string;
+  agent: string;
+  backend: AgentBackendKind;
+  /** Provider's own session id (resume token); null until the backend assigns one. */
+  agentSessionId: string | null;
+  /** opencode HTTP server port; null for ACP backends. */
+  serverPort: number | null;
+}
+
+/** Per-turn context passed to `prompt` (§2). */
+export interface PromptCtx {
+  worktreePath: string;
+  model: string;
+  signal: AbortSignal;
+  onEvent: (e: AgentEvent) => void;
+}
+
+/** Result of a completed turn (§2). Diff/persist happen outside the backend. */
+export interface TurnResult {
+  ok: boolean;
+  error?: string;
+}
+
+/**
+ * The core backend abstraction (§2). Implementations: OpenCodeServerBackend
+ * (Phase 1), WarmAcpBackend (Phase 2).
+ */
+export interface AgentBackend {
+  /** Lazy: spawn server / warm process if not already up for this (session, agent). §2 */
+  ensureSession(sessionId: string, opts: EnsureSessionOpts): Promise<AgentSessionHandle>;
+  /** Send a prompt; stream events via ctx.onEvent; resolves when the turn completes. §2 */
+  prompt(handle: AgentSessionHandle, input: string, ctx: PromptCtx): Promise<TurnResult>;
+  /** Graceful teardown of one session (session close or idle timeout). §2 */
+  closeSession(handle: AgentSessionHandle): Promise<void>;
+  /** Full teardown — kills all spawned servers/processes. §2 */
+  dispose(): Promise<void>;
+  /** Liveness for health endpoint + dispatcher fallback decision. §2 */
+  health(): 'up' | 'down';
+}

apps/coder/src/services/agent-pool.ts

@@ -0,0 +1,44 @@
+/**
+ * v2.6 — AgentPool (Phase 0 scaffold).
+ *
+ * Lazy get-or-create registry of `AgentBackend` instances keyed by
+ * `${sessionId}:${agent}`. Phase 0 ships the skeleton only: an in-memory Map,
+ * lookup / register / health, and clean disposal wired to the server's onClose.
+ * Spawning lands in Phase 1/2; nothing populates the map yet.
+ *
+ * Spec: openspec/changes/v2-6-persistent-agent-sessions/design.md §2.
+ */
+import type { AgentBackend } from './agent-backend.js';
+
+export class AgentPool {
+  private readonly backends = new Map<string, AgentBackend>();
+
+  private key(sessionId: string, agent: string): string {
+    return `${sessionId}:${agent}`;
+  }
+
+  /** Map lookup only. Spawning is Phase 1/2 — never creates here. */
+  get(sessionId: string, agent: string): AgentBackend | undefined {
+    return this.backends.get(this.key(sessionId, agent));
+  }
+
+  /** Store a backend instance for this (session, agent). */
+  register(sessionId: string, agent: string, backend: AgentBackend): void {
+    this.backends.set(this.key(sessionId, agent), backend);
+  }
+
+  /** Summary for the health endpoint. */
+  health(): { size: number } {
+    return { size: this.backends.size };
+  }
+
+  /** Dispose every backend and clear the map. Tolerates throwing backends. */
+  async dispose(): Promise<void> {
+    const entries = [...this.backends.values()];
+    this.backends.clear();
+    await Promise.allSettled(entries.map((b) => b.dispose()));
+  }
+}
+
+/** Single shared instance — referenced only by the server's onClose hook in Phase 0. */
+export const agentPool = new AgentPool();

data/coder-providers.example.json

@@ -0,0 +1,12 @@
+{
+  "providers": {
+    "goose": { "enabled": false },
+    "amp-acp": {
+      "extends": "acp",
+      "label": "Amp",
+      "description": "ACP wrapper for Amp",
+      "command": ["amp-acp"],
+      "enabled": true
+    }
+  }
+}

data/coder-providers.json

@@ -1,3 +0,0 @@
-{
-  "providers": {}
-}