Skip to content

armgit / mso

Public Imported from GitHub
0

Browser-based graphical shell and control plane for a Linux server you own — real terminal, file manager, live metrics and BYOK AI in one mobile-first workspace. Next.js 16 · React 19 · node-pty · Tailwind v4 · shadcn/ui. Public alpha, self-hosted.

fix(mcp): retain scanner limits and document mailbox parity

Abdurrahman Fakhrule4c5f5e9d99d
diff --git a/docs/CHANGELOG.md b/docs/CHANGELOG.md
index 2ff25575..a5d987bc 100644
--- a/docs/CHANGELOG.md
+++ b/docs/CHANGELOG.md
@@ -8,6 +8,7 @@ this is the *what*, and it is what Settings → Account → About shows as “Wh
**Fixed**
+- `mcp` retain scanner limits and document mailbox parity
- `security` close credential sinks and revoke pending sessions
- `mcp` preserve managed configuration behind credential guards
- `security` bind receipts to authority and pin browser artifacts
diff --git a/lib/mcp/parity.test.ts b/lib/mcp/parity.test.ts
index 4f9c39b8..081afe16 100644
--- a/lib/mcp/parity.test.ts
+++ b/lib/mcp/parity.test.ts
@@ -82,6 +82,7 @@ const MCP_ONLY: Record<string, string> = {
"local.agents.list": "native Local Agents address durable CLI/MCP AgentSession principals; Alfa browser threads use a separate thread lifecycle and do not own a compatible durable session receiver",
"local.agent.message.send": "same durable-session boundary; adding this to Alfa without first unifying its browser-thread identity would let one UI thread claim another principal's local receiver",
"local.agent.inbox": "the inbox is keyed to the exact durable AgentSession principal/id; Alfa currently has a separate owner-thread store rather than that receiver identity",
+ "local.agent.inbox.acknowledge": "acknowledgement changes the exact durable AgentSession mailbox; Alfa browser threads do not own that receiver identity",
"a2a.agents.list": "A2A is an external-agent interoperability surface for MCP/terminal agents; in-shell Alfa has no remote-agent registry yet",
"a2a.agent.discover": "public Agent Card discovery is intentionally available to the MCP/terminal agent harness and does not expose owner UI state",
"a2a.agent.register": "remote A2A registry mutation inherits MCP write approval/audit semantics; Alfa has no matching registry surface",
diff --git a/lib/mcp/tool-contract.ts b/lib/mcp/tool-contract.ts
index 7a07e6d2..2e1b90d8 100644
--- a/lib/mcp/tool-contract.ts
+++ b/lib/mcp/tool-contract.ts
@@ -135,7 +135,7 @@ export function toolDescriptor(tool: McpTool, profile: McpToolProfile = "full")
return {
name: tool.name,
title: tool.title ?? toolTitle(tool.name),
- description: compact ? compactText(tool.chatgptDescription ?? tool.description, 36) : tool.description,
+ description: compact ? compactText(tool.chatgptDescription ?? tool.description, 30) : tool.description,
inputSchema: compact ? compactSchema(tool.inputSchema) : tool.inputSchema,
...(outputSchema ? { outputSchema: compact ? compactSchema(outputSchema) : outputSchema } : {}),
securitySchemes,
diff --git a/lib/mcp/tools-local-agents.ts b/lib/mcp/tools-local-agents.ts
index c6273e9d..e89f5579 100644
--- a/lib/mcp/tools-local-agents.ts
+++ b/lib/mcp/tools-local-agents.ts
@@ -133,6 +133,7 @@ export const LOCAL_AGENT_TOOLS: McpTool[] = [
},
{
name: "local_agent_request_wait",
+ chatgptDescription: "Wait up to 30s for this session's exact correlated request reply; never resends or starts work.",
description: "Wait a bounded foreground interval for the correlated reply to one exact request previously sent by this session. Returns replied, target_offline, consumer_absent, or timeout with current receiver/standby observability. This never starts a background loop and never resends the request.",
scope: "read",
annotations: { readOnlyHint: true, idempotentHint: true },
@@ -177,6 +178,7 @@ export const LOCAL_AGENT_TOOLS: McpTool[] = [
},
{
name: "local_agent_inbox",
+ chatgptDescription: "Read this session's unread messages; wait_ms receives for up to 20s. Reply via local_agent_reply; acknowledge via write-scoped local_agent_inbox_acknowledge.",
description: "Read durable local-agent messages for this exact session. With wait_ms > 0, keep this foreground MCP call open for a bounded interval and return as soon as a peer message arrives; this legacy receiver remains separate from server-native standby. Each item carries explicit intent/correlation metadata. If intent=request, answer it with local_agent_reply. By default only unread items are returned.",
scope: "read",
annotations: { readOnlyHint: true, idempotentHint: true },
@@ -204,6 +206,7 @@ export const LOCAL_AGENT_TOOLS: McpTool[] = [
},
{
name:"local_agent_inbox_acknowledge",
+ chatgptDescription:"Mark exact inbox message_ids read for this session; requires write scope.",
description:"Mark exact messages in this conversation's inbox read. Requires write scope; read local_agent_inbox first and pass its exact message ids.",
scope:"write",
annotations:{readOnlyHint:false,idempotentHint:true,destructiveHint:false},
diff --git a/lib/mcp/tools-workflow-start.ts b/lib/mcp/tools-workflow-start.ts
index e6db4be1..fd519f98 100644
--- a/lib/mcp/tools-workflow-start.ts
+++ b/lib/mcp/tools-workflow-start.ts
@@ -19,7 +19,7 @@ import { prepareWorkflowStartWorkspace, startWorkflowWithWorkspace, workflowWork
export const WORKFLOW_START_TOOL: McpTool = {
name: "workflow_start",
description: "First call for multi-step work: resolve project context, search trusted skills/recipes/graphs, and return workflow_id. Carry that exact id on later steps. Read official skill mso-agent-bootstrap when learning the MSO map.",
- chatgptDescription: "First call for multi-step work; keep workflow_id.",
+ chatgptDescription: "First call for multi-step work",
scope: "write",
annotations: { idempotentHint: false },
outputSchema: WORKFLOW_START_OUTPUT,