How do I verify Montage workspace access with Kiro CLI?
Kiro CLI loads MCP servers through configuration files and agent-specific definitions, with a documented precedence hierarchy. It shares some remote configuration with Kiro IDE but differs in OAuth client-secret support and terminal interaction. This guide focuses on the CLI's effective agent configuration, avoiding legacy Amazon Q command examples and the incorrect assumption that a user-level entry must be the one an active agent actually loads.
Using Kiro CLI, call only Montage signals_status as the first live operation. Report the actual workspace identity and coverage from the tool response. If authentication, transport, or the tool is unavailable, stop and explain the blocker. Do not start collection, recurring tracking, writes, publishing, messaging, or social actions.Resolve the active agent’s configuration
Inspect the selected agent's mcpServers field, workspace .kiro/settings/mcp.json, and global ~/.kiro/settings/mcp.json. Kiro documents agent configuration above workspace configuration above global configuration. A disabled entry at a higher level can suppress a lower-level server. Identify the winning Montage definition before changing authentication, particularly when a custom agent was imported from another project.
Use a remote URL entry
Add or merge montage with url https://app.montages.co/mcp using the documented remote JSON shape. Do not reuse kiro-cli mcp add examples containing --command and --args for local servers. Leave wildcard autoApprove absent and review disabledTools if the agent already filters its catalog. Saving the right file matters more than registering the same address repeatedly under new names.
Review CLI-specific OAuth support
Kiro supports browser OAuth for self-configured remote servers in the CLI. Unlike the IDE's public-client restriction, the CLI also supports confidential client configuration when required by a service. That difference is not a reason to invent Montage client credentials. Use discovery and the supported consent path, and keep any existing secrets in approved configuration rather than terminal transcripts.
Inspect the loaded terminal server
In the interactive session, use /mcp to inspect loaded servers and confirm which Montage entry is active. Compare the actual tool names with the intended approval policy. The terminal dashboard shows discovery state, not private workspace authorization. If authentication fails, examine the documented OAuth error, including exact issuer matching, before modifying callbacks or scopes blindly.
Call status without unattended continuation
Ask the active CLI agent to execute Montage signals_status only and inspect the workspace returned by the server. Keep the terminal attended for the initial check; headless execution cannot be assumed to complete browser consent. Stop on failure and retain its category. No successful CLI connection authorizes automatic collection, recurring tasks, code-driven publishing, or social actions.
Keep the boundaries clear.
- CLI agent configuration overrides workspace then global MCP settings; Web/Mobile and IDE authentication limitations differ.
- Documented setup, not live interoperability certification. A visible tool catalog is not workspace authorization.
- First live operation: read-only signals_status. No automatic collection, recurring tracking, or social actions; broader actions need separate explicit authorization.
Start with a read-only check: ask your agent to report actual source coverage before collecting records or exporting an audience.
Primary references
Documentation reviewed 2026-11-15. Product capabilities, client availability and plans can change. Third-party marks identify their products, not an affiliation.