mulmoterminal 5.7.0 → 5.8.0

package/README.md

@@ -723,9 +723,9 @@ the `claude` / `codex` sessions themselves.
 | `CURSOR_BIN` | `cursor-agent` | The Cursor CLI binary to spawn. |
 | `CURSOR_MODEL` | cursor default | Model passed to Cursor as `--model` (unset = cursor's own default). Cursor's model names are **account-specific** and a wrong one is a hard exit — check `cursor-agent --list-models` first. |
 | `MULMOTERMINAL_HOME` | `~/.mulmoterminal` | Root for managed **git worktrees**. |
-| `CLAUDE_CONFIG_DIR` | `~` | Claude Code's own config directory. `.claude.json` lives **inside** it, so relocating your Claude Code config moves that file too — MulmoTerminal reads it to tell whether the per-project GUI MCP server is registered (`server/infra/gui-mcp-registration.ts`). Leave it unset and `~/.claude.json` is used. |
+| `CLAUDE_CONFIG_DIR` | `~/.claude` | Claude Code's own config directory. MulmoTerminal reads Claude's transcripts (`projects/`), prompt history (`history.jsonl`) and user skills (`skills/`) from it, and `.claude.json` from **inside** it (unset: `~/.claude.json`), which is how the per-project GUI MCP registration is checked (`server/session/project-dir.ts`). Set it in the environment MulmoTerminal starts from, so it matches the one Claude runs under. |
 | `MULMOCLAUDE_WORKSPACE_PATH` | `~/mulmoclaude` | Where the managed MulmoClaude workspace lives. MulmoTerminal seeds presets/helps **only** into this directory, so launching in an arbitrary project never writes them there (`server/backends/workspaceSetup.ts`), and it is what decides where MulmoTerminal's own runtime state goes — see the note under the table. Set it to the same value MulmoClaude uses. |
-| `MULMOTERMINAL_NO_SKILL_INSTALL` | unset | Set to any value to skip installing the bundled skills (`mulmoterminal-config` and the `-dirs` / `-theme` / `-header` / `-keys` / `-model` / `-notify` / `-bug-report` / `-decisions` family) into `~/.claude/skills/` and the Codex skills root on startup. |
+| `MULMOTERMINAL_NO_SKILL_INSTALL` | unset | Set to any value to skip installing the bundled skills (`mulmoterminal-config` and the `-dirs` / `-theme` / `-header` / `-keys` / `-model` / `-notify` / `-bug-report` / `-decisions` family) into Claude's skills root (`~/.claude/skills/`, and `$CLAUDE_CONFIG_DIR/skills/` as well when that is set) and the Codex skills root on startup. |
 | `GEMINI_IMAGE_MODEL` | `gemini-3.1-flash-image-preview` | Model used for image generation (needs `GEMINI_API_KEY`). The default is a **preview** model Google schedules for retirement around mid-2026, so pin a stable one here (e.g. `gemini-2.5-flash-image`) rather than waiting for a code change. |
 | `WAIT_REAP_GRACE_MS` | `1800000` | How long a **waiting** background session is kept before it's auto-reaped (`0` or negative = never). |
 
@@ -777,6 +777,7 @@ The Settings modal (the gear button) persists per-user UI choices to `~/.mulmote
 | `repoDirs`   | `{ "owner/repo": "/abs/path" }` — which local clone work on a repo starts in, when you keep several side by side. Only the *choice* is stored; which clones exist is re-derived from `cwdPresets` on every read, and an entry that no longer names a clone of that repo is ignored. |
 | `launchers`  | `{ label, command }` entries offered in a grid cell's launcher besides the agents — any interactive command. A plain shell needs no entry: the Agent Picker's **Shell** option opens `$SHELL` unconfigured. |
 | `customAgents` | `{ id, label, agent, command }` entries offered in the **Agent Picker** — your own way of starting Claude Code (`ollama launch claude --model … --`, a wrapper script). Unlike a launcher, Claude Code's own argv is **appended** to `command`, so the cell is a real session: resume, cost, context, GUI tools. `agent` says which agent's arguments to append and is required (`"claude"` is the only value today); `command` must stop taking arguments where Claude Code's begin — hence the trailing `--` above. Up to 8. |
+| `accounts` | **Beta** ([setup guide](docs/guide/en/accounts.md)). `{ id, label, agent, home }` entries — a second login for Claude Code (`agent: "claude"`) or Codex (`"codex"`), kept in its own config directory `home` (absolute or `~/…`). Picked per cell in the launch form's **ACCOUNT** select (the cell's header then names it); a session started on one runs with `CLAUDE_CONFIG_DIR` / `CODEX_HOME` set to it, and is bound to that account for good: resume, cost, titles and history are read from there, session lists show every account's rows, and the toolbar shows each account's 5h / 7d usage beside the default's. An account pointed at `~/.claude` / `~/.codex` itself is treated as the default login: no variable, no gauge of its own. A cell with no account gets no such variable added (whatever MulmoTerminal's own environment carries is inherited as before), so configuring none changes nothing. Up to 8. |
 | `quickCommands` | `{ label, text, agents? }` phrases the **phone** offers as chips on a session's terminal view. Tapping one puts `text` in the input box; it is not sent until you press send. `agents` scopes a chip to session kinds — any of `SESSION_AGENTS` (`"claude"`, `"codex"`, `"antigravity"`, `"grok"`, `"muse"`, `"copilot"`, `"cursor"`, `"shell"`); omit it to offer the chip everywhere. Empty by default. |
 | `userMcpServers` | `{ id, url }` HTTP MCP servers merged into the `--mcp-config` of the **Claude** sessions that carry the full GUI MCP (codex is handed the GUI server alone, `codexGuiMcpServers`) — a cell whose working directory is the **workspace**, and a session the server starts itself (the phone, a scheduled task) unless it asks for a grid cell's shape, as an issue's seed session does (`issueSpawnOptions`). A cell in a project directory does not get this merge; the MCP config the user wrote is read either way. Takes effect on the next session. |
 | `buttons`    | Header action buttons — see [Header buttons](#header-buttons). Omit to keep the defaults; set to replace them. |
@@ -1782,7 +1783,7 @@ newest first, including freshly-created sessions that aren't yet written to disk
 }
 ```
 
-- Sessions are read from `~/.claude/projects/<encoded CLAUDE_CWD>/*.jsonl` and
+- Sessions are read from `<claude config home>/projects/<encoded CLAUDE_CWD>/*.jsonl` (`CLAUDE_CONFIG_DIR`, default `~/.claude`) and
   merged with in-memory sessions started this run but not yet persisted (those
   have `title: "New session"` and `mtime` = creation time).
 - Sorted by `mtime` descending and capped at the **50** most recent. Files are
@@ -1955,7 +1956,8 @@ same-origin-guarded.
 | `POST /api/worktrees/create` · `/remove` · `/push` · `/pr` | Create on `agent/<slug>` — or, with `issue: <N>`, on `issue/<N>-<slug>` forked from a freshly fetched `origin/<base>`; remove (managed root only), push, open a PR (`gh`, else compare URL). |
 | `GET /api/prs` · `GET /api/issues` | Open PRs / issues across the configured `prRepos` — `gh` for github.com entries, `glab` for gitlab.com and any host declared in `gitlabHosts`. |
 | `GET /api/repo-dirs` | Which saved directories clone which GitHub repo, ordered, with the recorded choice per repo. |
-| `POST /api/issues/start` | Cut an issue's worktree in one of that repo's known clones and spawn a session there, seeded with the issue as a draft. |
+| `POST /api/issues/start` | Cut an issue's worktree in one of that repo's known clones and spawn a session there, seeded with the issue. An optional `agent` (any hosted agent; absent is `claude`, anything else is a 400) picks which. A Claude seed is a draft; every other agent runs it at once. The reply names the `agent` and whether the seed runs (`seedRuns`), so the cell attaches on the right endpoint. |
+| `GET /api/agents/availability` | Which hosted agents this machine can start — `{ agents: [{ agent, available, reason?, installGuide? }] }`, `reason` being `missing`, `no-such-path` (an `<AGENT>_BIN` naming nothing) or `not-executable`, and `installGuide` the agent's official install page (from `bin/agent-install-guides.json`) or `null`. The same check a spawn makes before it starts, run once at server start, so an agent installed later shows after a restart. |
 | `GET /api/github/star` · `POST /api/github/star` | Whether you have starred MulmoTerminal, and star it (via `gh`). `starred: null` means `gh` could not answer, and hides the button. |
 
 **Workspace views**
@@ -2259,7 +2261,7 @@ Codex sessions are unaffected — the CLI has no equivalent flag.
 ## Session discovery & titles
 
 Claude stores each project's sessions as JSONL files under
-`~/.claude/projects/<encoded-cwd>/<session-id>.jsonl`, where the absolute `cwd`
+`~/.claude/projects/<encoded-cwd>/<session-id>.jsonl` (under `CLAUDE_CONFIG_DIR` instead of `~/.claude` when that is set), where the absolute `cwd`
 has its `/` and `.` characters replaced with `-` (e.g.
 `/Users/you/proj` → `-Users-you-proj`).
 

package/bin/agent-install-guides.d.ts

@@ -0,0 +1,10 @@
+// Types for bin/agent-install-guides.js — hand-written for the reason every other bin/*.d.ts is: the
+// file is plain JS because it runs before tsx does, and the specs and the server are TypeScript.
+export interface AgentInstallGuideEntry {
+  url: string;
+  /** The date (YYYY-MM-DD) the page was last opened and confirmed to be this agent's install guide. */
+  checked: string;
+}
+
+export declare function agentInstallGuide(agent: string): string | null;
+export declare const AGENT_INSTALL_GUIDES: Readonly<Record<string, AgentInstallGuideEntry>>;

package/bin/agent-install-guides.js

@@ -0,0 +1,23 @@
+// Where each hosted agent's OFFICIAL install guide lives (#2230). The URLs themselves are data, in
+// agent-install-guides.json beside this file — edit that, not this.
+//
+// Links only, never install commands, for agent-bins.js's reason: a wrong command is the one thing
+// a stuck user will paste. Each entry is a page on the maker's own domain that gives the install
+// steps for the command this app runs; `checked` is the date someone last opened it and confirmed
+// that. An agent whose page cannot be confirmed is left OUT rather than guessed — its readers get
+// no link, which is better than a wrong one.
+//
+// Here in bin/ because the CLI's missing-agent message runs before tsx exists; the server imports
+// it from here (as it does agent-bins.js), and the browser reads it off /api/agents/availability.
+import { createRequire } from "node:module";
+
+const GUIDES = createRequire(import.meta.url)("./agent-install-guides.json");
+
+/** The install guide for `agent`, or null when none is recorded. */
+export function agentInstallGuide(agent) {
+  const entry = Object.prototype.hasOwnProperty.call(GUIDES, agent) ? GUIDES[agent] : null;
+  return entry !== null && typeof entry === "object" && typeof entry.url === "string" ? entry.url : null;
+}
+
+/** The raw table, for the spec that checks every entry's shape. */
+export const AGENT_INSTALL_GUIDES = GUIDES;

package/bin/agent-install-guides.json

@@ -0,0 +1,9 @@
+{
+  "claude": { "url": "https://code.claude.com/docs/en/setup", "checked": "2026-09-24" },
+  "codex": { "url": "https://learn.chatgpt.com/docs/codex/cli", "checked": "2026-09-24" },
+  "antigravity": { "url": "https://antigravity.google/docs/cli/install/", "checked": "2026-09-24" },
+  "grok": { "url": "https://docs.x.ai/build/overview", "checked": "2026-09-24" },
+  "muse": { "url": "https://dev.meta.ai/docs/muse-code", "checked": "2026-09-24" },
+  "copilot": { "url": "https://docs.github.com/en/copilot/how-tos/copilot-cli/set-up-copilot-cli/install-copilot-cli", "checked": "2026-09-24" },
+  "cursor": { "url": "https://cursor.com/docs/cli/installation", "checked": "2026-09-24" }
+}

package/bin/default-agent.js

@@ -13,6 +13,7 @@
 // saved Claude cell as something else. This value is read when a NEW session starts, and nowhere
 // in the path that restores an old one.
 import { AGENT_BIN_SPEC, AGENT_INSTALL_HINT } from "./agent-bins.js";
+import { agentInstallGuide } from "./agent-install-guides.js";
 
 export const isKnownAgent = (value) => typeof value === "string" && Object.prototype.hasOwnProperty.call(AGENT_BIN_SPEC, value);
 
@@ -92,6 +93,10 @@ export function missingAgentMessage(gate, bin) {
     lines.push('  ~/.mulmoterminal/config.json:  { "defaultAgent": "codex" }');
     lines.push(`  agents: ${Object.keys(AGENT_BIN_SPEC).join(", ")}`);
   }
+  // The maker's own install page (#2230), from the same table the Agent Picker links to — so the
+  // terminal and the browser cannot send someone to two different places for one missing agent.
+  const install = agentInstallGuide(gate.agent);
+  if (install) lines.push(`Install guide:  ${install}`);
   lines.push("");
   lines.push(`Details:  ${GUIDE.en}`);
   lines.push(`日本語:    ${GUIDE.ja}`);

package/common/agentAccounts.ts

@@ -0,0 +1,61 @@
+// An ACCOUNT: a second login for an agent CLI, kept in its own config home (#2215).
+//
+// Claude Code and Codex each keep their login, transcripts and settings in one directory, and
+// both let the environment move it (CLAUDE_CONFIG_DIR / CODEX_HOME). An account names such a
+// directory, so a cell can run on a different subscription from the cell beside it.
+//
+// It is NOT a custom agent (common/customAgents.ts). A custom agent is HOW the CLI is started;
+// an account is WHOSE login it runs under. The two combine: an `ollama launch claude` custom
+// agent can run on the work account.
+//
+// Configuring none changes nothing. A cell without an account has no variable ADDED to its spawn —
+// not even the default value, because Claude Code keys its keychain entry on the variable being
+// SET: pointing CLAUDE_CONFIG_DIR at `~/.claude` explicitly is already a different login. What the
+// server's own environment already carries is inherited as before.
+import { isRecord } from "./isRecord.js";
+
+// The CLIs whose home an account can move. Each needs the spawn to set its variable and every
+// reader of its state to follow the session's home, so adding one is that work, not a label.
+export const ACCOUNT_AGENTS = ["claude", "codex"] as const;
+
+export type AccountAgent = (typeof ACCOUNT_AGENTS)[number];
+
+export const isAccountAgent = (value: unknown): value is AccountAgent => ACCOUNT_AGENTS.some((agent) => agent === value);
+
+export interface AgentAccount {
+  /** Stable slug. It keys the wire (`?account=`) and the per-session record, so renaming the
+   *  LABEL is free while renaming this strands the sessions started on it. */
+  id: string;
+  /** What the launch form and the cell show. */
+  label: string;
+  /** Which CLI this is a login for. */
+  agent: AccountAgent;
+  /** The config home, absolute or `~/`-relative, as written. */
+  home: string;
+}
+
+// Lowercase slug, for the same reasons as a custom agent id: it travels in a query string and is
+// compared exactly on both sides.
+export const ACCOUNT_ID_RE = /^[a-z0-9][a-z0-9_-]{0,31}$/;
+
+export const isAccountId = (value: unknown): value is string => typeof value === "string" && ACCOUNT_ID_RE.test(value);
+
+// Relative homes are refused rather than resolved: the CLI would resolve one against each cell's
+// own working directory, i.e. a different home — and login — per directory.
+export const isAccountHome = (value: unknown): value is string =>
+  typeof value === "string" && (value.startsWith("/") || value.startsWith("~/") || /^[A-Za-z]:[\\/]/.test(value));
+
+/** A wire/config row read back as an AgentAccount. Used by the server sanitizing config.json and
+ *  by the browser filtering GET /api/config. */
+export function isAgentAccount(row: unknown): row is AgentAccount {
+  if (!isRecord(row)) return false;
+  return isAccountId(row.id) && typeof row.label === "string" && !!row.label.trim() && isAccountAgent(row.agent) && isAccountHome(row.home);
+}
+
+/** The accounts one agent can run on, in config order — none for an agent an account cannot move. */
+export const accountsForAgent = (accounts: readonly AgentAccount[], agent: unknown): AgentAccount[] =>
+  isAccountAgent(agent) ? accounts.filter((account) => account.agent === agent) : [];
+
+/** What a cell or a list row calls an account: its label, or the bare id once the entry has left the
+ *  config (the session still runs on it — see server/session/account-log.ts). */
+export const accountLabel = (accounts: readonly AgentAccount[], id: string): string => accounts.find((account) => account.id === id)?.label ?? id;

package/common/agentAvailability.ts

@@ -0,0 +1,50 @@
+// Which agents this machine can start, as GET /api/agents/availability reports it (#2229). Here
+// rather than in server/ because the launch pickers will read it to grey out an agent that cannot
+// run, and #2230 words its install guidance from `reason`.
+import { isRecord } from "./isRecord";
+import { isTerminalAgent, type TerminalAgent } from "./sessionAgent";
+
+/** Why an agent cannot be started — the kind of the spawn preflight's own diagnosis, so the advice
+ *  can differ: `missing` is not installed (or not on the server's PATH), `no-such-path` is an
+ *  `<AGENT>_BIN` override naming nothing, `not-executable` is a file that cannot run. */
+export const AGENT_UNAVAILABLE_REASONS = ["missing", "no-such-path", "not-executable"] as const;
+
+export type AgentUnavailableReason = (typeof AGENT_UNAVAILABLE_REASONS)[number];
+
+// No `bin`: with an `<AGENT>_BIN` override it is a path on this machine, and nothing that reads
+// this needs it — the reason already says whether an override is what failed.
+export type AgentAvailability =
+  | { agent: TerminalAgent; available: true }
+  /** `installGuide` is the agent's official install page, or null when none is recorded (#2230). */
+  | { agent: TerminalAgent; available: false; reason: AgentUnavailableReason; installGuide: string | null };
+
+export interface AgentAvailabilityResponse {
+  agents: AgentAvailability[];
+}
+
+const isUnavailableReason = (value: unknown): value is AgentUnavailableReason => AGENT_UNAVAILABLE_REASONS.some((reason) => reason === value);
+
+function parseEntry(raw: unknown): AgentAvailability | null {
+  if (!isRecord(raw) || typeof raw.agent !== "string" || !isTerminalAgent(raw.agent)) return null;
+  if (raw.available === true) return { agent: raw.agent, available: true };
+  if (raw.available !== false || !isUnavailableReason(raw.reason)) return null;
+  return { agent: raw.agent, available: false, reason: raw.reason, installGuide: httpsUrlOrNull(raw.installGuide) };
+}
+
+// Only an https link is ever rendered: this goes into an href, and the server is the one source,
+// but the page must not be the thing that trusts it.
+function httpsUrlOrNull(value: unknown): string | null {
+  if (typeof value !== "string") return null;
+  try {
+    return new URL(value).protocol === "https:" ? value : null;
+  } catch {
+    return null;
+  }
+}
+
+/** The response body as the browser receives it, entry by entry: a malformed entry is dropped, not
+ *  the whole answer, and an agent with no entry reads as available — today's behaviour. */
+export function parseAgentAvailabilityResponse(raw: unknown): AgentAvailability[] {
+  if (!isRecord(raw) || !Array.isArray(raw.agents)) return [];
+  return raw.agents.map(parseEntry).filter((entry): entry is AgentAvailability => entry !== null);
+}

package/dist/assets/{abnfDiagram-O67JEVCF-CYOfx7Yo.js → abnfDiagram-O67JEVCF-D_iilXNw.js}

@@ -1 +1 @@
-import{m as e,t}from"./mermaid-parser.core-Co2Ckrb2.js";import{n}from"./chunk-Y2CYZVJY-DsF7k-Jl.js";import{m as r}from"./src-CllyRBe0.js";import"./chunk-O7XYJQB3-BCjP9Rta.js";import{n as i,r as a,t as o}from"./chunk-GTNCS2PH-GehvsYOa.js";import{t as s}from"./chunk-JWPE2WC7-DVXcaiue.js";import"./mermaid.core-D_YUclq0.js";var c=e().RailroadAbnf.parser.LangiumParser,l=n(e=>{let t=e.alternatives.map(u);return t.length===1?t[0]:{type:`choice`,alternatives:t}},`transformAlternation`),u=n(e=>{let t=e.elements.map(f);return t.length===1?t[0]:{type:`sequence`,elements:t}},`transformConcatenation`),d=n(e=>{if(e.includes(`*`)){let[t,n]=e.split(`*`);return{min:t?parseInt(t,10):0,max:n?parseInt(n,10):1/0}}let t=parseInt(e,10);return{min:t,max:t}},`parseRepeat`),f=n(e=>{let t=p(e.primary);if(!e.repeat)return t;let{min:n,max:r}=d(e.repeat);return n===0&&r===1?{type:`optional`,element:t}:{type:`repetition`,element:t,min:n,max:r}},`transformElement`),p=n(e=>{switch(e.$type){case`AbnfStringLiteral`:return{type:`terminal`,value:e.value};case`AbnfNumVal`:return{type:`terminal`,value:e.value};case`AbnfRuleName`:return{type:`nonterminal`,name:e.name};case`AbnfGroup`:return l(e.element);case`AbnfOptionalGroup`:return{type:`optional`,element:l(e.element)};default:throw Error(`Unsupported ABNF primary node: ${e.$type}`)}},`transformPrimary`),m=n(e=>({name:e.name,definition:l(e.definition)}),`transformRule`),h=n(e=>{s(e,o),e.title&&o.setTitle(e.title),e.rules.map(e=>o.addRule(m(e)))},`populateDb`),g={parser:{parse:n(e=>{o.clear(),r.debug(`[ABNF Parser] Starting Langium parse`);let n=c.parse(e);if(n.lexerErrors.length>0||n.parserErrors.length>0)throw new t(n);let i=n.value;r.debug(`[ABNF Parser] Parsed rules:`,i.rules.length),h(i),r.debug(`[ABNF Parser] Parse complete`)},`parse`),parser:{yy:o}},db:o,renderer:a,styles:i};export{g as diagram};
+import{m as e,t}from"./mermaid-parser.core-C9FaUx4X.js";import{n}from"./chunk-Y2CYZVJY-DsF7k-Jl.js";import{m as r}from"./src-CllyRBe0.js";import"./chunk-O7XYJQB3-CIi6yVqu.js";import{n as i,r as a,t as o}from"./chunk-GTNCS2PH-DNCVJdqM.js";import{t as s}from"./chunk-JWPE2WC7-DVXcaiue.js";import"./mermaid.core-C86wCgRX.js";var c=e().RailroadAbnf.parser.LangiumParser,l=n(e=>{let t=e.alternatives.map(u);return t.length===1?t[0]:{type:`choice`,alternatives:t}},`transformAlternation`),u=n(e=>{let t=e.elements.map(f);return t.length===1?t[0]:{type:`sequence`,elements:t}},`transformConcatenation`),d=n(e=>{if(e.includes(`*`)){let[t,n]=e.split(`*`);return{min:t?parseInt(t,10):0,max:n?parseInt(n,10):1/0}}let t=parseInt(e,10);return{min:t,max:t}},`parseRepeat`),f=n(e=>{let t=p(e.primary);if(!e.repeat)return t;let{min:n,max:r}=d(e.repeat);return n===0&&r===1?{type:`optional`,element:t}:{type:`repetition`,element:t,min:n,max:r}},`transformElement`),p=n(e=>{switch(e.$type){case`AbnfStringLiteral`:return{type:`terminal`,value:e.value};case`AbnfNumVal`:return{type:`terminal`,value:e.value};case`AbnfRuleName`:return{type:`nonterminal`,name:e.name};case`AbnfGroup`:return l(e.element);case`AbnfOptionalGroup`:return{type:`optional`,element:l(e.element)};default:throw Error(`Unsupported ABNF primary node: ${e.$type}`)}},`transformPrimary`),m=n(e=>({name:e.name,definition:l(e.definition)}),`transformRule`),h=n(e=>{s(e,o),e.title&&o.setTitle(e.title),e.rules.map(e=>o.addRule(m(e)))},`populateDb`),g={parser:{parse:n(e=>{o.clear(),r.debug(`[ABNF Parser] Starting Langium parse`);let n=c.parse(e);if(n.lexerErrors.length>0||n.parserErrors.length>0)throw new t(n);let i=n.value;r.debug(`[ABNF Parser] Parsed rules:`,i.rules.length),h(i),r.debug(`[ABNF Parser] Parse complete`)},`parse`),parser:{yy:o}},db:o,renderer:a,styles:i};export{g as diagram};

package/dist/assets/architecture-7GRP2DOG-Dg93bZQr.js

@@ -0,0 +1 @@
+import"./chunk-FOHPRMQF-DzxwRta7.js";import{M as e}from"./mermaid-parser.core-C9FaUx4X.js";export{e as createArchitectureServices};