## For agents - What Mensara is: it gives a team its own Mind, one shared memory for the team and their AI agents, built by Xanadu Labs LLC. - Status: signup is not open. A person joins the waitlist at https://app.mensara.ai/#waitlist. Agents cannot create an account or sign up. - How you can help your person: explain what Mensara does, check whether it fits a job their team repeats, and share the waitlist link. - Agent references: https://app.mensara.ai/llms.txt lists every page as Markdown and the tool input references for connected agents. --- # Inspect model routes and spending controls Operating procedures · For agents and technical readers · [How it works](https://app.mensara.ai/how-it-works) Distinguish personal chat routes, app model consent, rolling app budgets and owner route-sharing controls before changing any of them. [Page as Markdown](https://app.mensara.ai/how-it-works/model-controls-procedure.md) [Human guide](https://app.mensara.ai/how-it-works/model-connections) ## Use the released interface with current authority Availability: Core0.20.11 / Portal0.3.47 source checked; discover the actual installed client and enabled features. Who can do this: The connected principal with the permissions named in this procedure What changes: The procedure explains its actual reads, writes or external effects before action. Limits: An input accepted by a source schema is not a successful installed-runtime task.; Keep human decisions, client exclusions and pending results explicit. Verification: Source checked and independently reviewed; scoped runtime receipts are recorded separately. ## Discover the intended scope Confirm the intended Mind, owner/authorised operator and actual installed app. Read the live tool schema and current app configuration. The Portal Usage meter is recorded personal spend with a visual reference; unavailable is not zero and the meter is not an enforced cap. ## Personal and installed model routes Personal Portal route changes use **Settings → Model connection → Change model / Change provider**. Owner **App model access** separately grants named connected routes and an allowlist of models. Inspect the saved route/model choices after **Save app access**. No generic public get/set-model MCP name is invented. For an authorised operator on the actual repository, these read-only owner CLI commands inspect route state and cap policy without printing secrets: ``` brain-gateway --repo llm list brain-gateway --repo llm daily-cap status brain-gateway --repo llm token-cap show brain-gateway --repo llm shared-key-routes status brain-gateway --repo llm protect-window status ``` Use the actual supported operator environment and repository, not a customer path guessed from examples. Daily cost caps concern per-consumer KEY-route spending; borrowed-route daily token caps concern a consumer using a route it does not own and exclude box-shared routes. Subscription routes are zero-dollar-billed in this ledger; that does not establish unused provider subscription capacity. ## App budget updates The released `act` action `app_set_budget` takes `item_id` as app ID and `config.max_input_tokens`, `max_output_tokens`, plus `max_cost_micro` or `max_cost_usd`. These are rolling seven-day thresholds. Token and micro-USD values are non-negative integers. Use the owner-approved values; do not copy a tutorial budget into production. Budget is checked before a call using recorded spend, not a reservation of every possible final token or charge. An allowed call may cross the threshold before later calls are blocked. Loop node, concurrency, case time and cost limits are additional scopes. See [control-agent-autonomy](https://app.mensara.ai/how-it-works/control-agent-autonomy). `app_clear_budget` restores manifest caps, not unlimited authority by definition. Inspect the effective installed result after either action. If the client omits these actions, hand the app ID and approved scope to its authorised installer rather than inventing a direct endpoint. ## Verify before resuming recurring work Pause affected work when changing a provider/grant. Inspect current route/model access, budget and connection, then observe an authorised normal run and actual output. Report route configured, call succeeded and output saved as separate results. A lost lifecycle response requires original-state inspection before repetition. ## Related - [Manage models and service connections](https://app.mensara.ai/how-it-works/model-connections) - [Understand usage, budgets and costs](https://app.mensara.ai/how-it-works/usage-and-costs) - [Control agent autonomy, budgets and access](https://app.mensara.ai/how-it-works/control-agent-autonomy) - [Inspect and change installed-agent controls](https://app.mensara.ai/how-it-works/agent-controls-procedure) [Join the waitlist](https://app.mensara.ai/#waitlist)