MCP v2 PR-3: write_pane, spawn_pane, connect_host + SSH safeguards

Three of the highest-power v2 tools, plus a defense-in-depth pass
on SSH-specific risk.

write_pane sends keystrokes (or any bytes) to a pane's PTY. The
policy engine matches against the text content directly so rules
like write_pane(npm test*) match by what would run, and the
compiled-in hard-deny catches rm -rf /, fork bombs, etc. regardless
of policy. Per-pane token-bucket rate limiter (30 calls / 10s,
3/sec refill) prevents a runaway loop from spamming the user with
confirm modals or burning audit-log capacity. The frontend handler
truncates the text in modal/audit summaries to ~60 chars + escapes
control characters so secrets pasted into write_pane don't echo
verbatim into the UI.

spawn_pane mirrors the existing SpawnSpec enum (WSL distro,
PowerShell, SSH) as the tool schema. New splitLeafWith helper
inserts a caller-built LeafNode (with a pre-generated id) so the
handler can await waitForPaneRegistration on that exact leaf before
replying with the resulting {leafId, paneId}. 15s spawn timeout
covers cold-start WSL distros; 30s for connect_host covers SSH
handshake + auth. Outer dispatch timeout bumped 30s → 60s. SSH
spawns without a saved hostId are refused — LeafNode only persists
sshHostId, no inline params, so use connect_host.

connect_host is a thin wrapper that looks up a saved SSH host by
id and routes through the same spawn machinery.

McpConfirm.tsx gains an optional ssh context — when the call
targets or spawns an SSH pane, a red warning banner renders
explaining that pattern matching is best-effort on the bytes we
send (remote shell expands aliases/subshells before executing).
buildConfirmSummary became buildConfirmInfo and returns the SSH
context alongside the summary string.

PR-3.5 — SSH safeguards. Two new switches in the Policy tab,
both off by default, both gated by mcp_policy::SshSafeguards:

  allowOpenSsh: when off, connect_host and spawn_pane(kind=ssh)
    refuse server-side with a clear "ssh-disabled" message pointing
    at the Policy tab. User must open SSH manually via the titlebar
    🔑 picker and toggle 🤖 on to grant Claude access.

  autoAllowSpawnedSsh: when off, an SSH pane Claude spawns starts
    with mcpAllow=false. User must explicitly toggle 🤖 before
    Claude can read scrollback or send keystrokes. The second switch
    is disabled in the UI when the first is off.

The safe-by-default design means a fresh install gives Claude no
ability to autonomously touch SSH — full safety with one click per
level to enable when consciously wanted. Both switches read fresh
per call so policy edits take effect without a server restart.

ErrorBoundary.tsx — last-resort guard against React render
exceptions. Wraps the App root + each MCP panel tab independently
so a bug in one tab doesn't blank the entire app. Shows a small
red error card with the exception message and a "Try again"
button. Caught a serde rename_all bug during PR-3.5 testing where
PolicyTab read policy.sshSafeguards but Rust serialized
ssh_safeguards (snake_case); without the boundary the whole window
went black.

newId() now exported from tree.ts for the splitLeafWith path.
McpPolicy struct gained #[serde(rename_all = "camelCase")] so
sshSafeguards survives the IPC round-trip cleanly; older policy
files without the field still load (serde defaults to safe).
This commit is contained in:
megaproxy 2026-05-26 14:50:06 +01:00
parent 3acad63fb7
commit bf2810a433
12 changed files with 844 additions and 41 deletions

View file

@ -237,6 +237,45 @@ fn truncate_summary(s: &str) -> String {
}
}
// ----------------------------------------------------------------------------
// Write rate limiter (per OWASP LLM06 "Excessive Agency" + MCP spec MUST).
// Token bucket per leaf_id, only applied to write_pane.
// ----------------------------------------------------------------------------
const WRITE_RATE_CAPACITY: f64 = 30.0;
/// Tokens added per second. 30 / 10s = 3.0 → burst of 30, sustained 3/s.
const WRITE_RATE_REFILL_PER_SEC: f64 = 3.0;
#[derive(Default)]
pub struct WriteRateLimiter {
/// leaf_id → (last_refill, tokens). Naive HashMap that never evicts; for
/// a long-running session with many transient panes this could grow, but
/// LeafIds are uuid-shaped and a few hundred entries is fine.
buckets: PlMutex<HashMap<String, (Instant, f64)>>,
}
impl WriteRateLimiter {
/// Try to consume a token for this leaf. Returns Ok on success, Err with
/// the milliseconds to wait until the next token will be available.
pub fn try_consume(&self, leaf_id: &str) -> Result<(), u64> {
let mut buckets = self.buckets.lock();
let now = Instant::now();
let entry = buckets
.entry(leaf_id.to_string())
.or_insert((now, WRITE_RATE_CAPACITY));
let elapsed = now.saturating_duration_since(entry.0).as_secs_f64();
entry.1 = (entry.1 + elapsed * WRITE_RATE_REFILL_PER_SEC).min(WRITE_RATE_CAPACITY);
entry.0 = now;
if entry.1 >= 1.0 {
entry.1 -= 1.0;
Ok(())
} else {
let wait_secs = (1.0 - entry.1) / WRITE_RATE_REFILL_PER_SEC;
Err((wait_secs * 1000.0).ceil() as u64)
}
}
}
// ----------------------------------------------------------------------------
// MCP service: tools + resources.
// ----------------------------------------------------------------------------
@ -246,6 +285,7 @@ pub struct TileService {
ptys: Arc<PtyManager>,
state: Arc<RwLock<McpState>>,
pending: Arc<PendingActions>,
rate_limiter: Arc<WriteRateLimiter>,
app: AppHandle,
tool_router: ToolRouter<Self>,
}
@ -326,6 +366,51 @@ pub struct ApplyPresetArgs {
pub allow_drops: bool,
}
#[derive(Debug, Deserialize, schemars::JsonSchema)]
pub struct SpawnPaneArgs {
/// What shell to run in the new pane: WSL distro, PowerShell, or a
/// saved SSH host (or arbitrary ssh args).
pub spec: crate::pty::SpawnSpec,
/// Where to insert the new pane. Defaults to the active pane, or the
/// workspace root if no pane is active. The parent leaf must be
/// MCP-allowed for Claude to target it.
#[serde(default)]
pub parent_leaf_id: Option<LeafId>,
/// "h" → new pane to the right; "v" → new pane below. If omitted,
/// picks based on the parent's current aspect ratio (wider → "h",
/// taller → "v"), matching the titlebar "+" button's behaviour.
#[serde(default)]
pub orientation: Option<String>,
}
#[derive(Debug, Deserialize, schemars::JsonSchema)]
pub struct ConnectHostArgs {
/// Stable id of a host saved via the SSH host manager. The new pane
/// inherits the host's user/port/identity-file/etc. and reuses the
/// keyring-stored password (if any) for auto-fill at the SSH prompt.
pub host_id: String,
/// Same semantics as spawn_pane.parent_leaf_id.
#[serde(default)]
pub parent_leaf_id: Option<LeafId>,
/// Same semantics as spawn_pane.orientation.
#[serde(default)]
pub orientation: Option<String>,
}
#[derive(Debug, Deserialize, schemars::JsonSchema)]
pub struct WritePaneArgs {
/// Stable leaf id from the tree (uuid-shaped). Must belong to a pane
/// the user has allow-listed for MCP access.
pub leaf_id: LeafId,
/// Bytes to send to the pane's PTY. Use "\n" for Enter. Each call sends
/// one chunk; partial commands are fine but block the shell until you
/// send a newline. This is the highest-risk MCP tool — Claude can send
/// arbitrary keystrokes including destructive commands. The policy
/// engine + hard-deny list are evaluated against this text directly.
/// Rate-limited to 30 calls / 10s per pane.
pub text: String,
}
#[derive(Debug, Deserialize, schemars::JsonSchema)]
pub struct SetLabelArgs {
/// Stable leaf id from the tree (uuid-shaped). Must belong to a pane
@ -344,12 +429,14 @@ impl TileService {
ptys: Arc<PtyManager>,
state: Arc<RwLock<McpState>>,
pending: Arc<PendingActions>,
rate_limiter: Arc<WriteRateLimiter>,
app: AppHandle,
) -> Self {
Self {
ptys,
state,
pending,
rate_limiter,
app,
tool_router: Self::tool_router(),
}
@ -466,7 +553,11 @@ impl TileService {
let _ = self.app.emit("mcp://request", &payload);
// 6. Await reply with 30s timeout.
let result = tokio::time::timeout(Duration::from_secs(30), rx).await;
// 60s is the outer cap. Per-tool inner timeouts (waitForPaneRegistration
// in the frontend handler) are tighter for fast ops and looser for
// SSH/spawn — this just keeps a misbehaving frontend from leaking a
// request id forever.
let result = tokio::time::timeout(Duration::from_secs(60), rx).await;
let duration_ms = now_ms() - start_ms;
@ -664,6 +755,169 @@ impl TileService {
}
}
#[tool(description = "Send keystrokes (text) to a pane's PTY. The leaf \
must be MCP-allowed. Use \"\\n\" for Enter. Rate-limited per pane: \
30 calls per 10 seconds. The user policy is evaluated against the \
sent text (so rules like write_pane(git push *) match by content), \
and a compiled-in hard-deny list always catches rm -rf /, fork \
bombs, mkfs on devices, etc. those are blocked regardless of \
policy.")]
async fn write_pane(
&self,
Parameters(args): Parameters<WritePaneArgs>,
) -> Result<CallToolResult, McpError> {
let _leaf = self.require_visible_leaf(&args.leaf_id).await?;
// Rate limit BEFORE dispatch — we don't want a misbehaving client
// to spam the user with confirm modals or burn audit-log capacity.
if let Err(retry_after_ms) = self.rate_limiter.try_consume(&args.leaf_id) {
tracing::warn!(
leaf_id = %args.leaf_id,
retry_after_ms,
"write_pane: rate limited"
);
return Err(McpError::invalid_params(
format!("rate limited; retry after {retry_after_ms}ms"),
Some(json!({
"leaf_id": &args.leaf_id,
"retry_after_ms": retry_after_ms,
})),
));
}
// args_repr IS the text — that's what hard-deny and the user's
// policy globs pattern-match against. (For other tools we use a
// stable summary string; for write_pane the text is the surface.)
let args_repr = args.text.clone();
let args_json = json!({ "leafId": &args.leaf_id, "text": &args.text });
tracing::debug!(
leaf_id = %args.leaf_id,
bytes = args.text.len(),
"write_pane: dispatching"
);
let _ = self
.dispatch_action("write_pane", args_json, args_repr)
.await?;
Ok(CallToolResult::success(vec![Content::text("ok")]))
}
#[tool(description = "Spawn a new pane next to an existing one, running \
the requested shell (WSL distro / PowerShell / SSH). The new pane is \
auto-allowed for MCP so Claude can immediately read its scrollback \
and send it keystrokes (subject to policy). Returns {leafId, paneId} \
for the newly created pane. Times out after ~15 seconds if the \
PTY can't be spawned (covers a cold WSL distro start).")]
async fn spawn_pane(
&self,
Parameters(args): Parameters<SpawnPaneArgs>,
) -> Result<CallToolResult, McpError> {
// SSH safeguard: refuse before any other work if the user hasn't
// opted in to MCP-initiated SSH connections.
if matches!(args.spec, crate::pty::SpawnSpec::Ssh { .. })
&& !self.policy_ssh_open_allowed().await
{
return Err(McpError::invalid_params(
"ssh-disabled: Claude is not allowed to open SSH connections \
(Policy tab SSH safeguards 'Allow Claude to open SSH \
connections'). Open the SSH session manually via the \
titlebar 🔑 picker first, then ask Claude to interact with \
it.",
None,
));
}
// If a parent is named, it must be MCP-allowed. (If omitted, the
// frontend picks the active pane or root — no explicit check here
// since the user already approved an MCP-server-running state.)
if let Some(ref parent) = args.parent_leaf_id {
let _ = self.require_visible_leaf(parent).await?;
}
// Serialise the spec back to JSON so the frontend handler can
// reconstruct it without us doing the per-kind dispatch here.
let spec_json = serde_json::to_value(&args.spec).map_err(|e| {
McpError::internal_error(
format!("serialize SpawnSpec: {e}"),
None,
)
})?;
let kind_str = match &args.spec {
crate::pty::SpawnSpec::Wsl { .. } => "wsl",
crate::pty::SpawnSpec::Powershell => "powershell",
crate::pty::SpawnSpec::Ssh { .. } => "ssh",
};
let args_repr = format!(
"shell={} parent={} orientation={}",
kind_str,
args.parent_leaf_id.as_deref().unwrap_or("(active)"),
args.orientation.as_deref().unwrap_or("(auto)"),
);
let args_json = json!({
"spec": spec_json,
"parentLeafId": args.parent_leaf_id,
"orientation": args.orientation,
});
tracing::debug!(shell = kind_str, "spawn_pane: dispatching");
let result = self
.dispatch_action("spawn_pane", args_json, args_repr)
.await?;
Ok(CallToolResult::success(vec![Content::text(
result.to_string(),
)]))
}
#[tool(description = "Open a new pane connected to a saved SSH host. \
Thin wrapper around spawn_pane that resolves host_id against the \
saved-hosts list (use the tiletopia://hosts resource to list them). \
Returns {leafId, paneId} for the new pane. The user's saved password \
for the host (if any) is auto-typed at the prompt passwords are \
never exposed through the MCP surface.")]
async fn connect_host(
&self,
Parameters(args): Parameters<ConnectHostArgs>,
) -> Result<CallToolResult, McpError> {
if !self.policy_ssh_open_allowed().await {
return Err(McpError::invalid_params(
"ssh-disabled: Claude is not allowed to open SSH connections \
(Policy tab SSH safeguards 'Allow Claude to open SSH \
connections'). Open the SSH session manually via the \
titlebar 🔑 picker first, then ask Claude to interact with \
it.",
None,
));
}
if let Some(ref parent) = args.parent_leaf_id {
let _ = self.require_visible_leaf(parent).await?;
}
// Verify host_id is in the mirror (so Claude can't probe unknown ids).
let host_known = {
let st = self.state.read().await;
st.mirror.hosts.iter().any(|h| h.id == args.host_id)
};
if !host_known {
return Err(McpError::invalid_params(
"unknown host_id (use tiletopia://hosts to list saved hosts)",
Some(json!({ "host_id": &args.host_id })),
));
}
let args_repr = format!(
"host={} parent={} orientation={}",
&args.host_id,
args.parent_leaf_id.as_deref().unwrap_or("(active)"),
args.orientation.as_deref().unwrap_or("(auto)"),
);
let args_json = json!({
"hostId": &args.host_id,
"parentLeafId": args.parent_leaf_id,
"orientation": args.orientation,
});
tracing::debug!(host_id = %args.host_id, "connect_host: dispatching");
let result = self
.dispatch_action("connect_host", args_json, args_repr)
.await?;
Ok(CallToolResult::success(vec![Content::text(
result.to_string(),
)]))
}
#[tool(description = "Set or clear the human-readable label on a pane. \
Pass empty string to clear. The leaf must be MCP-allowed.")]
async fn set_label(
@ -782,6 +1036,20 @@ impl TileService {
Ok(CallToolResult::success(vec![Content::text("ok")]))
}
/// Read the persisted SSH-safeguard switch. Fresh-read every call so a
/// user editing the policy in the panel takes effect on the next MCP
/// call without a server restart. Errors fall back to the safe default
/// (refuse).
async fn policy_ssh_open_allowed(&self) -> bool {
match crate::mcp_policy::load_or_init(&self.app) {
Ok(p) => p.ssh_safeguards.allow_open_ssh,
Err(e) => {
tracing::warn!(error = %e, "policy_ssh_open_allowed: load failed, defaulting to false");
false
}
}
}
/// Shared validation for tools that target an existing leaf — confirms
/// the leaf is in the mirror (which means the user has it allow-listed
/// for MCP) and returns its metadata. Factored out of the 4+ tools that
@ -968,6 +1236,9 @@ pub async fn start_server(
let ptys_f = ptys.clone();
let state_f = state.clone();
let pending_f = pending.clone();
// Single shared rate limiter for the lifetime of this server — token
// buckets are per-leaf-id, but the registry itself is one piece of state.
let rate_limiter: Arc<WriteRateLimiter> = Arc::new(WriteRateLimiter::default());
// Clone AppHandle before the move closure so we can pass it into each
// TileService instance. AppHandle is cheap to clone (it's an Arc inside).
let app_handle_for_service = app_handle.clone();
@ -982,6 +1253,7 @@ pub async fn start_server(
ptys_f.clone(),
state_f.clone(),
pending_f.clone(),
rate_limiter.clone(),
app_handle_for_service.clone(),
))
},

View file

@ -13,9 +13,34 @@ use tauri::{AppHandle, Manager};
// ---------------------------------------------------------------------------
#[derive(Clone, Debug, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct McpPolicy {
pub version: u32, // currently 1
pub permissions: McpPermissions,
/// SSH-specific safety toggles. Both default to false ("safest") on
/// fresh installs and when loading older policy files that pre-date
/// this section. Surface in the Policy tab's "SSH safeguards" section.
#[serde(default)]
pub ssh_safeguards: SshSafeguards,
}
#[derive(Clone, Debug, Default, Serialize, Deserialize)]
#[serde(rename_all = "camelCase")]
pub struct SshSafeguards {
/// When false (default), the connect_host and spawn_pane(kind=ssh)
/// MCP tools refuse server-side with a clear error — Claude can't
/// initiate SSH connections at all; the user must open them manually
/// via the titlebar 🔑 picker. Turn on if you want Claude to be able
/// to open saved SSH hosts on its own (e.g. for multi-host work).
#[serde(default)]
pub allow_open_ssh: bool,
/// When false (default), an SSH pane that Claude spawns starts with
/// mcpAllow=false — Claude can't read its scrollback or send it
/// keystrokes until the user explicitly toggles 🤖 on. Turn on if
/// you want full autonomy once you've consented to the open-ssh
/// switch above.
#[serde(default)]
pub auto_allow_spawned_ssh: bool,
}
#[derive(Clone, Debug, Default, Serialize, Deserialize)]
@ -329,6 +354,7 @@ pub fn default_policy() -> McpPolicy {
McpPolicy {
version: 1,
permissions: McpPermissions::default(),
ssh_safeguards: SshSafeguards::default(),
}
}

View file

@ -21,7 +21,10 @@ pub type PaneId = u64;
/// Discriminated union describing what to spawn into a fresh PTY. Serialized
/// as `{ kind: "wsl" | "powershell" | "ssh", ... }` from the frontend.
#[derive(Debug, Clone, Deserialize)]
/// Also reused as the schema for the MCP `spawn_pane` tool — `JsonSchema`
/// lets rmcp render it for Claude; `Serialize` lets the backend bounce it
/// back into the `mcp://request` event payload for the frontend handler.
#[derive(Debug, Clone, Deserialize, Serialize, schemars::JsonSchema)]
#[serde(tag = "kind", rename_all = "lowercase")]
pub enum SpawnSpec {
Wsl {

View file

@ -19,6 +19,7 @@ import {
writeToPane,
killPane,
type PaneId,
type SpawnSpec,
type SshHost,
type McpStatus,
type McpMirror,
@ -35,7 +36,9 @@ import {
type LeafNode,
type LeafShellSpec,
newLeaf,
newId,
splitLeaf,
splitLeafWith,
closeLeaf,
findLeaf,
leafCount,
@ -94,6 +97,21 @@ function defaultShellAsLeafProps(d: DefaultShell): Partial<LeafNode> {
return { shellKind: "wsl", distro: d.distro };
}
/** Cap a string for display in modals / audit summaries. Single-line, max
* 60 visible chars, control characters escaped so secrets pasted into a
* write_pane call don't print as gibberish in the modal. */
function truncateForSummary(s: string, cap = 60): string {
const oneLine = s.replace(/\r?\n/g, "\\n").replace(/\t/g, "\\t");
return oneLine.length > cap ? oneLine.slice(0, cap) + "…" : oneLine;
}
/** Short human-readable form of a SpawnSpec, used in MCP confirm summaries. */
function describeSpec(spec: SpawnSpec): string {
if (spec.kind === "wsl") return `WSL${spec.distro ? ` (${spec.distro})` : ""}`;
if (spec.kind === "powershell") return "PowerShell";
return `SSH${spec.host ? ` to ${spec.host}` : ""}`;
}
export default function App() {
// ---- top-level state -----------------------------------------------------
const [tree, setTree] = useState<TreeNode>(() => newLeaf());
@ -577,14 +595,134 @@ export default function App() {
return () => window.removeEventListener("keydown", onKey, true);
}, [split, close, toggleBroadcast, promoteActive]);
// Waiters keyed by leaf id — used by the MCP spawn_pane / connect_host
// handlers, which must reply with the new paneId but can only get one
// after the freshly-mounted XtermPane completes its spawn round-trip and
// calls back into registerPaneId.
const pendingPaneRegistrations = useRef<Map<NodeId, (paneId: PaneId) => void>>(
new Map(),
);
const registerPaneId = useCallback(
(leafId: NodeId, paneId: PaneId | null) => {
if (paneId == null) paneIdByLeafRef.current.delete(leafId);
else paneIdByLeafRef.current.set(leafId, paneId);
if (paneId == null) {
paneIdByLeafRef.current.delete(leafId);
return;
}
paneIdByLeafRef.current.set(leafId, paneId);
const waiter = pendingPaneRegistrations.current.get(leafId);
if (waiter) {
pendingPaneRegistrations.current.delete(leafId);
waiter(paneId);
}
},
[],
);
/** Insert a new leaf into the tree from a SpawnSpec used by the MCP
* spawn_pane and connect_host handlers. Returns the new leaf's id
* (caller awaits waitForPaneRegistration on it for the paneId).
* - parentLeafId: defaults to active leaf, then first leaf in layout.
* - orientation: "h" / "v" / undefined; undefined picks smart-orient by
* parent pane aspect (matches the titlebar "+" button's behaviour).
* - New leaf is auto-marked mcpAllow=true so Claude can immediately
* interact with the pane it just spawned. */
const spawnNewLeafFromSpec = useCallback(
async (
spec: SpawnSpec,
parentLeafId: string | undefined,
orientationArg: string | undefined,
): Promise<NodeId> => {
const layout = flattenLayout(treeRef.current);
const parentId =
parentLeafId ?? activeLeafId ?? layout.leaves[0]?.leaf.id ?? null;
if (!parentId) throw new Error("no pane available to split off");
const parent = findLeaf(treeRef.current, parentId);
if (!parent || parent.kind !== "leaf") {
throw new Error(`parent leaf not found: ${parentId}`);
}
let orient: Orientation;
if (orientationArg === "h" || orientationArg === "v") {
orient = orientationArg;
} else if (orientationArg == null) {
// Smart-orient: split along the longer side of the parent's pane.
const container = paneWrapRef.current;
const slot = layout.leaves.find((s) => s.leaf.id === parentId);
if (container && slot) {
const rect = container.getBoundingClientRect();
const paneW = slot.box.width * rect.width;
const paneH = slot.box.height * rect.height;
orient = paneW >= paneH ? "h" : "v";
} else {
orient = "h"; // safe fallback
}
} else {
throw new Error(
`invalid orientation: ${orientationArg} (expected "h" or "v")`,
);
}
// For SSH spawns, the auto-allow safeguard decides whether the new
// pane starts MCP-allowed (Claude can interact immediately) or
// mcpAllow=off (user must explicitly toggle 🤖 to grant access).
// Local shells (WSL / PowerShell) are auto-allowed unconditionally.
let mcpAllow = true;
if (spec.kind === "ssh") {
try {
const policy = await mcpPolicyLoad();
mcpAllow = policy.sshSafeguards.autoAllowSpawnedSsh;
} catch (e) {
console.warn("policy load failed during ssh spawn, defaulting mcpAllow=false:", e);
mcpAllow = false;
}
}
const id = newId();
const newLeafNode: LeafNode = {
kind: "leaf",
id,
shellKind: spec.kind,
mcpAllow,
...(spec.kind === "wsl"
? { distro: spec.distro, cwd: spec.cwd }
: {}),
...(spec.kind === "ssh" && spec.hostId
? { sshHostId: spec.hostId }
: {}),
};
setTree((t) => splitLeafWith(t, parentId, orient, newLeafNode));
return id;
},
[activeLeafId],
);
/** Resolves to the paneId once XtermPane finishes mounting and the
* spawn_pane Tauri command returns. Rejects after timeoutMs. */
function waitForPaneRegistration(
leafId: NodeId,
timeoutMs = 5000,
): Promise<PaneId> {
return new Promise((resolve, reject) => {
const existing = paneIdByLeafRef.current.get(leafId);
if (existing != null) {
resolve(existing);
return;
}
pendingPaneRegistrations.current.set(leafId, resolve);
setTimeout(() => {
if (pendingPaneRegistrations.current.has(leafId)) {
pendingPaneRegistrations.current.delete(leafId);
reject(
new Error(
`spawn timed out after ${timeoutMs}ms — pane never registered`,
),
);
}
}, timeoutMs);
});
}
const broadcastFrom = useCallback(
(originLeafId: NodeId, dataB64: string) => {
let peers = 0;
@ -829,6 +967,81 @@ export default function App() {
summary: `Rename pane "${before}" → "${after}"`,
};
}
case "write_pane": {
const a = args as { leafId?: string; text?: string };
if (typeof a.leafId !== "string") throw new Error("missing leafId");
if (typeof a.text !== "string") throw new Error("missing text");
const leaf = findLeaf(treeRef.current, a.leafId);
if (!leaf || leaf.kind !== "leaf") throw new Error(`leaf not found: ${a.leafId}`);
const paneId = paneIdByLeafRef.current.get(a.leafId);
if (paneId == null) throw new Error(`no live pane for leaf ${a.leafId}`);
// UTF-8 encode → base64 (matches the existing writeToPane wire shape).
const bytes = new TextEncoder().encode(a.text);
let binary = "";
for (let i = 0; i < bytes.byteLength; i++) {
binary += String.fromCharCode(bytes[i]);
}
const b64 = btoa(binary);
await writeToPane(paneId, b64);
const labelStr = leaf.label ?? a.leafId.slice(0, 8);
return {
payload: { leafId: a.leafId, bytesWritten: bytes.length },
summary: `Send to pane "${labelStr}": ${truncateForSummary(a.text)}`,
};
}
case "spawn_pane": {
const a = args as {
spec?: SpawnSpec;
parentLeafId?: string;
orientation?: string;
};
if (!a.spec || typeof a.spec !== "object") {
throw new Error("missing spec");
}
// For SSH, our LeafNode only persists sshHostId — we don't store
// inline host/user/port on the leaf. Refuse ad-hoc SSH spawns;
// use connect_host with a saved host_id instead.
if (a.spec.kind === "ssh" && !a.spec.hostId) {
throw new Error(
"spawn_pane with kind=ssh requires hostId (a saved host). " +
"Use connect_host(host_id) or add the host via the SSH host manager first.",
);
}
const newLeafId = await spawnNewLeafFromSpec(
a.spec,
a.parentLeafId,
a.orientation,
);
// 15s covers a cold WSL distro start (~3-5s typical, longer if
// the distro hasn't been used recently).
const paneId = await waitForPaneRegistration(newLeafId, 15000);
return {
payload: { leafId: newLeafId, paneId },
summary: `Spawn ${describeSpec(a.spec)} pane`,
};
}
case "connect_host": {
const a = args as {
hostId?: string;
parentLeafId?: string;
orientation?: string;
};
if (typeof a.hostId !== "string") throw new Error("missing hostId");
const host = hosts.find((h) => h.id === a.hostId);
if (!host) throw new Error(`unknown host_id: ${a.hostId}`);
const newLeafId = await spawnNewLeafFromSpec(
{ kind: "ssh", host: host.hostname, hostId: host.id },
a.parentLeafId,
a.orientation,
);
// 30s covers SSH handshake + password auth on a slow or
// first-time connection.
const paneId = await waitForPaneRegistration(newLeafId, 30000);
return {
payload: { leafId: newLeafId, paneId, hostId: host.id },
summary: `Connect SSH to "${host.label}" (${host.hostname})`,
};
}
case "close_pane": {
const a = args as { leafId?: string };
if (typeof a.leafId !== "string") throw new Error("missing leafId");
@ -931,45 +1144,93 @@ export default function App() {
throw new Error(`unsupported MCP tool: ${tool}`);
}
},
[setLabel, close, defaultShell, activeLeafId],
[setLabel, close, defaultShell, activeLeafId, hosts, spawnNewLeafFromSpec],
);
// The summary string for the confirm modal needs access to the leaf
// metadata, so we compute it up-front by partially running the handler
// logic (without mutating). For now we just rebuild it inline per tool;
// when more tools land this should split out.
const buildConfirmSummary = useCallback((tool: string, args: unknown): string => {
function leafLabel(id: string | undefined): string {
if (!id) return "(unknown)";
const l = findLeaf(treeRef.current, id);
return l && l.kind === "leaf" ? (l.label ?? id.slice(0, 8)) : id.slice(0, 8);
}
switch (tool) {
case "set_label": {
const a = args as { leafId?: string; label?: string };
return `Rename pane "${leafLabel(a.leafId)}" → "${a.label || "(cleared)"}"`;
const buildConfirmInfo = useCallback(
(tool: string, args: unknown): { summary: string; ssh?: { hostLabel: string } } => {
function leafLabel(id: string | undefined): string {
if (!id) return "(unknown)";
const l = findLeaf(treeRef.current, id);
return l && l.kind === "leaf" ? (l.label ?? id.slice(0, 8)) : id.slice(0, 8);
}
case "close_pane": {
const a = args as { leafId?: string };
return `Close pane "${leafLabel(a.leafId)}"`;
function sshContextForLeaf(id: string | undefined): { hostLabel: string } | undefined {
if (!id) return undefined;
const l = findLeaf(treeRef.current, id);
if (!l || l.kind !== "leaf" || l.shellKind !== "ssh") return undefined;
const host = hosts.find((h) => h.id === l.sshHostId);
return { hostLabel: host ? `${host.label} (${host.hostname})` : (l.sshHostId ?? "?") };
}
case "swap_panes": {
const a = args as { leafA?: string; leafB?: string };
return `Swap panes "${leafLabel(a.leafA)}" ↔ "${leafLabel(a.leafB)}"`;
switch (tool) {
case "set_label": {
const a = args as { leafId?: string; label?: string };
return {
summary: `Rename pane "${leafLabel(a.leafId)}" → "${a.label || "(cleared)"}"`,
};
}
case "write_pane": {
const a = args as { leafId?: string; text?: string };
return {
summary: `Send to pane "${leafLabel(a.leafId)}": ${truncateForSummary(a.text ?? "")}`,
ssh: sshContextForLeaf(a.leafId),
};
}
case "spawn_pane": {
const a = args as { spec?: SpawnSpec };
const summary = a.spec ? `Spawn ${describeSpec(a.spec)} pane` : "Spawn pane";
const ssh =
a.spec && a.spec.kind === "ssh"
? {
hostLabel: a.spec.hostId
? hosts.find((h) => h.id === a.spec!.hostId)?.label ?? a.spec.host
: a.spec.host,
}
: undefined;
return { summary, ssh };
}
case "connect_host": {
const a = args as { hostId?: string };
const host = a.hostId ? hosts.find((h) => h.id === a.hostId) : null;
const name = host ? `"${host.label}" (${host.hostname})` : a.hostId;
return {
summary: `Connect SSH to ${name}`,
ssh: host ? { hostLabel: `${host.label} (${host.hostname})` } : undefined,
};
}
case "close_pane": {
const a = args as { leafId?: string };
return {
summary: `Close pane "${leafLabel(a.leafId)}"`,
ssh: sshContextForLeaf(a.leafId),
};
}
case "swap_panes": {
const a = args as { leafA?: string; leafB?: string };
return {
summary: `Swap panes "${leafLabel(a.leafA)}" ↔ "${leafLabel(a.leafB)}"`,
};
}
case "promote_pane": {
const a = args as { leafId?: string };
return {
summary: `Promote pane "${leafLabel(a.leafId)}" up one level`,
};
}
case "apply_preset": {
const a = args as { name?: string; allowDrops?: boolean };
const suffix = a.allowDrops ? " (drops allowed)" : "";
return { summary: `Reshape workspace to ${a.name}${suffix}` };
}
default:
return { summary: `Run ${tool}` };
}
case "promote_pane": {
const a = args as { leafId?: string };
return `Promote pane "${leafLabel(a.leafId)}" up one level`;
}
case "apply_preset": {
const a = args as { name?: string; allowDrops?: boolean };
const suffix = a.allowDrops ? " (drops allowed)" : "";
return `Reshape workspace to ${a.name}${suffix}`;
}
default:
return `Run ${tool}`;
}
}, []);
},
[hosts],
);
useEffect(() => {
let cancelled = false;
@ -977,12 +1238,13 @@ export default function App() {
void onMcpRequest(async (req: McpActionRequest) => {
try {
if (req.needsConfirm) {
const summary = buildConfirmSummary(req.tool, req.args);
const info = buildConfirmInfo(req.tool, req.args);
const ok = await requestConfirm({
tool: req.tool,
args: req.args,
reason: req.reason,
summary,
summary: info.summary,
ssh: info.ssh,
});
if (!ok) {
await mcpActionReply(req.requestId, { Err: "user rejected" });
@ -1002,7 +1264,7 @@ export default function App() {
cancelled = true;
if (unlisten) unlisten();
};
}, [runMcpHandler, requestConfirm, buildConfirmSummary]);
}, [runMcpHandler, requestConfirm, buildConfirmInfo]);
const applyPreset = useCallback(
(make: (d: Partial<LeafNode>) => TreeNode) => {

View file

@ -0,0 +1,84 @@
import { Component, type ReactNode } from "react";
interface Props {
children: ReactNode;
/** Optional label for the error message ("Policy tab", "Audit log", etc.). */
label?: string;
}
interface State {
error: Error | null;
}
/** Last-resort guard against React render exceptions. Without this, a single
* bad render in any component blanks the entire app react unmounts the
* whole tree because the exception bubbles past the root. Wrap the App
* body or individual high-risk components (PolicyTab, AuditTab) with this. */
export default class ErrorBoundary extends Component<Props, State> {
state: State = { error: null };
static getDerivedStateFromError(error: Error): State {
return { error };
}
componentDidCatch(error: Error, info: { componentStack?: string | null }) {
// Surface to dev tools console — Tauri's WebView2 will show this in
// its inspector. Keeps the diagnostic accessible even if the panel
// refuses to render.
console.error("[ErrorBoundary]", this.props.label ?? "(unlabelled)", error, info);
}
handleReset = () => {
this.setState({ error: null });
};
render() {
if (this.state.error) {
return (
<div
style={{
padding: 14,
margin: 10,
background: "#1a0e0e",
border: "1px solid #6a2a2a",
borderRadius: 4,
color: "#e0a0a0",
font: "12px/1.5 monospace",
}}
role="alert"
>
<div style={{ fontWeight: 600, color: "#ff8080", marginBottom: 6 }}>
{this.props.label ?? "Component"} crashed while rendering
</div>
<pre
style={{
whiteSpace: "pre-wrap",
wordBreak: "break-word",
margin: "6px 0",
color: "#c08080",
fontSize: 11,
}}
>
{this.state.error.message}
</pre>
<button
onClick={this.handleReset}
style={{
marginTop: 6,
font: "inherit",
background: "#2a1a1a",
color: "#e0a0a0",
border: "1px solid #6a2a2a",
borderRadius: 3,
padding: "3px 10px",
cursor: "pointer",
}}
>
Try again
</button>
</div>
);
}
return this.props.children;
}
}

View file

@ -7,6 +7,11 @@ export interface McpConfirmSpec {
/** Human-readable summary of what's about to happen, computed by the
* per-tool handler (e.g. "rename pane 'shell' to 'build'"). */
summary: string;
/** Set when the action targets (or spawns) an SSH-connected pane. The
* modal renders an extra warning banner SSH targets bypass our
* in-app safety net since the remote shell expands aliases/subshells
* before executing, and the policy engine only sees the bytes we send. */
ssh?: { hostLabel: string };
}
interface McpConfirmProps {
@ -49,6 +54,16 @@ export default function McpConfirm({ spec, onAccept, onReject, onAlwaysAllow }:
</span>
</header>
<div className="mcp-confirm-body">
{spec.ssh && (
<div className="mcp-confirm-ssh-warn">
<strong>SSH target extra caveats apply.</strong>{" "}
This runs on the remote host <code>{spec.ssh.hostLabel}</code>.
The pattern matching in your policy only sees the bytes
tiletopia sends; the remote shell expands aliases, subshells,
and variables before executing. The hard-deny list still
applies, but treat this as <em>best-effort</em>, not a sandbox.
</div>
)}
<p className="mcp-confirm-summary">{spec.summary}</p>
{spec.reason && (
<p className="mcp-confirm-reason">

View file

@ -710,3 +710,68 @@
color: #ccd;
border-color: #4488cc;
}
.mcp-confirm-ssh-warn {
background: #2a1a1a;
border: 1px solid #a04040;
border-radius: 4px;
padding: 8px 10px;
margin: 0 0 10px;
color: #e0a0a0;
font-size: 11px;
line-height: 1.5;
}
.mcp-confirm-ssh-warn strong { color: #ff8080; }
.mcp-confirm-ssh-warn code {
background: #0c0c0c;
padding: 1px 4px;
border-radius: 2px;
color: #ffcccc;
}
.mcp-confirm-ssh-warn em { color: #ffd0a0; font-style: normal; }
/* ---- SSH safeguards section ------------------------------------------- */
.policy-ssh-safeguards {
background: #1a1410;
border: 1px solid #4a2a1a;
border-radius: 4px;
padding: 10px 12px;
margin-bottom: 12px;
}
.policy-ssh-safeguards .policy-bucket-header {
color: #d8a040;
border-bottom-color: #3a2a1a;
margin-bottom: 8px;
}
.policy-toggle-row {
display: flex;
align-items: flex-start;
gap: 8px;
padding: 6px 0;
cursor: pointer;
border-top: 1px solid #2a1a10;
}
.policy-toggle-row:first-of-type { border-top: none; }
.policy-toggle-row input[type="checkbox"] {
margin-top: 3px;
accent-color: #d8a040;
flex-shrink: 0;
}
.policy-toggle-text {
font-size: 11px;
color: #b8a890;
line-height: 1.45;
}
.policy-toggle-text strong { color: #d8a040; display: block; margin-bottom: 2px; }
.policy-toggle-text code {
background: #0c0c0c;
padding: 1px 4px;
border-radius: 2px;
font-family: inherit;
color: #ffcc80;
}
.policy-toggle-row input:disabled + .policy-toggle-text {
opacity: 0.5;
}

View file

@ -5,6 +5,7 @@ import {
import type { McpStatus, McpAuditEntry } from "../ipc";
import AuditTab from "./AuditTab";
import PolicyTab from "./PolicyTab";
import ErrorBoundary from "./ErrorBoundary";
import "./McpPanel.css";
interface McpPanelProps {
@ -294,10 +295,16 @@ export default function McpPanel({
)}
{tab === "audit" && (
<AuditTab entries={auditEntries} onClear={onClearAudit} />
<ErrorBoundary label="Audit tab">
<AuditTab entries={auditEntries} onClear={onClearAudit} />
</ErrorBoundary>
)}
{tab === "policy" && <PolicyTab />}
{tab === "policy" && (
<ErrorBoundary label="Policy tab">
<PolicyTab />
</ErrorBoundary>
)}
</div>
</div>
</>

View file

@ -126,6 +126,16 @@ export default function PolicyTab() {
}));
}
function setSshSafeguard(
key: "allowOpenSsh" | "autoAllowSpawnedSsh",
value: boolean,
) {
mutate((p) => ({
...p,
sshSafeguards: { ...p.sshSafeguards, [key]: value },
}));
}
async function handleSave() {
if (!policy || !dirty || saving) return;
setSaving(true);
@ -166,6 +176,40 @@ export default function PolicyTab() {
</div>
</div>
<div className="policy-ssh-safeguards">
<div className="policy-bucket-header">SSH safeguards</div>
<label className="policy-toggle-row">
<input
type="checkbox"
checked={policy.sshSafeguards.allowOpenSsh}
onChange={(e) => setSshSafeguard("allowOpenSsh", e.target.checked)}
/>
<div className="policy-toggle-text">
<strong>Allow Claude to open SSH connections.</strong> When off,
the <code>connect_host</code> and <code>spawn_pane(kind=ssh)</code>
{" "}tools refuse with a clear error. You can still open SSH
sessions manually via the titlebar 🔑 picker, and Claude can
interact with them if you toggle 🤖 on.
</div>
</label>
<label className="policy-toggle-row">
<input
type="checkbox"
checked={policy.sshSafeguards.autoAllowSpawnedSsh}
onChange={(e) =>
setSshSafeguard("autoAllowSpawnedSsh", e.target.checked)
}
disabled={!policy.sshSafeguards.allowOpenSsh}
/>
<div className="policy-toggle-text">
<strong>Auto-grant Claude access to newly-spawned SSH panes.</strong>{" "}
When off, an SSH pane Claude opens starts with 🤖 off you have
to explicitly toggle it before Claude can read scrollback or send
keystrokes. Only meaningful when the switch above is on.
</div>
</label>
</div>
<div className="policy-buckets">
{(["deny", "ask", "allow"] as Bucket[]).map((bucket) => (
<RuleList

View file

@ -165,6 +165,12 @@ export interface McpPolicy {
ask: string[];
allow: string[];
};
/** SSH-specific capability switches; mirrors Rust SshSafeguards. Both
* default to false on first load. */
sshSafeguards: {
allowOpenSsh: boolean;
autoAllowSpawnedSsh: boolean;
};
}
export const mcpPolicyLoad = (): Promise<McpPolicy> =>

View file

@ -71,7 +71,7 @@ export interface SplitNode {
export type TreeNode = LeafNode | SplitNode;
function newId(): NodeId {
export function newId(): NodeId {
return (
globalThis.crypto?.randomUUID?.() ??
Math.random().toString(36).slice(2, 12)
@ -158,6 +158,22 @@ export function splitLeaf(
});
}
/** Like {@link splitLeaf} but inserts a caller-constructed LeafNode (with a
* predetermined id) rather than minting a fresh one. Used by the MCP
* spawn_pane handler which needs the id up-front so it can wait for the
* matching registerPaneId call before replying to the backend. */
export function splitLeafWith(
root: TreeNode,
leafId: NodeId,
orientation: Orientation,
leaf: LeafNode,
): TreeNode {
return replaceById(root, leafId, (node) => {
if (node.kind !== "leaf") return node;
return newSplit(orientation, node, leaf);
});
}
/**
* Remove the leaf with the given id. The other child of its parent split
* takes the parent's place in the tree. Returns null if the closed leaf

View file

@ -2,12 +2,15 @@ import { StrictMode } from "react";
import { createRoot } from "react-dom/client";
import "./styles.css";
import App from "./App";
import ErrorBoundary from "./components/ErrorBoundary";
const root = document.getElementById("root");
if (!root) throw new Error("No #root element found");
createRoot(root).render(
<StrictMode>
<App />
<ErrorBoundary label="tiletopia">
<App />
</ErrorBoundary>
</StrictMode>
);