The model proposes. The host decides.

Host code outside the model answers what an agent asks: an approval gate, not an operating-system jail.

Read what it does not enforce

Where the boundary is

Two ways to read one boundary: who decides what runs, and where your data goes. Each label opens its row in the table under the map.

Who decides what runs: the model proposes, host code checks, you answer what it asks, and what you approve runs as you.

  1. The modelProposes a tool call. Nothing it says is a decision.
  2. Checked in codeNarrower than it sounds
  3. The host gateDeny-list, task mode and the run's grant, checked in code.
  4. Only you answerNarrower than it sounds
  5. YouAnswer what your approval mode sends you.
  6. Runs as youNarrower than it sounds
  7. What runsRuns with your user's reach, or in the container or VM tier you chose.
  8. An HMAC per entryNarrower than it sounds
  9. The recordThe security log, with an HMAC over each entry.
Holds
True at this release, with a live consumer rather than only a declaration.
Narrower than it sounds
Something real ships, but the obvious reading of the claim is wider than what this release does. The row states the narrower truth.
Every boundary, with its evidence and its known limit
BoundaryResponsibilityEnforcementEvidenceKnown limit
The model's proposal to the host gateAuthorityA tool call is only a proposal. Host code checks the deny-list, the task mode and the run's grant before any approval is consulted, so not even the most permissive approval mode can run what a task mode refuses.Narrower than it soundsagents/native/runtime.py, _guard_and_invokeAn agent CLI meets these checks only for the calls it asks about.
The host gate to youAuthorityYour approval mode decides which calls wait for you, and only you answer them: an agent's tool or an app cannot approve what waits on you.Narrower than it soundsapproval_answer.py, refusalPer-tool grants bind subagents, workflow leaves and unwatched turns, not the agent you chat with.
Your decision to what runsAuthorityAn approved command runs with your user's filesystem and network reach, with the gateway's secrets left out of its environment. A container or virtual-machine tier is opt-in.Narrower than it soundssandbox.py, build_child_envThe agent's shell is screened, not fenced, from your credential files.
What runs to the record of itAuthorityEach security event is written with an HMAC over its own record, so an entry edited in place fails the integrity check.Narrower than it soundssel.py, verify_integrityA deleted entry goes unreported: the check allows gaps, because several processes write one log.
Your network to the gatewayNetworkThe gateway asks for a local token by default. The one mode without it binds to loopback, so turning auth off cannot open the gateway to your network.Holdsauth/modes.py, effective_bindA process running as you can answer as you.
The gateway to an appNetworkAn app reaches the gateway with a token scoped to that app, through the gateway's own proxy, and an undeclared write is refused.Narrower than it soundsdashboard/token_auth.py, app_session_tokenAn app's own code runs as you, and its backend runs on the host unless its manifest names a sandbox tier.
The gateway to the internetNetworkCore's fetches, the model clients and an app's SDK requests pass one guard that refuses private, loopback and link-local addresses and dials the address it checked.Narrower than it soundsnet/guard.py, evaluateBy default it is not a destination allow-list, and an app backend's own traffic does not pass it.
The gateway to the provider you choseNetworkWhat a prompt is built from is masked before the model is handed it: a key, a token or a URL's login is found by its shape or its field's name.Narrower than it soundssecurity.py, redact_for_modelA secret written as a plain word reaches the model as written, and an agent CLI's own tools are outside the mask.

What it does not enforce

Read against v0.3.9, the last release we verified, on November 16, 2026. Each one links where it is stated in full.

Read every limitation

The controls, and the code that enforces each

Each links the file that enforces it at v0.3.9. A control is listed only if that code has a live caller at the tag; a setting nothing reads is not a control.

  • Task modes

    A tool the task mode does not allow is refused before any approval is asked, so an auto-approve cannot run it.

    agents/native/runtime.py, _guard_and_invoke
  • Approval modes

    Your approval mode decides which tool calls wait for you. Auto-approve expires, and turning it off fires every surface's disable.

    trust_mode.py
  • Only you answer

    An ask that waits on you is answered by you: a request an agent's tool or an app sends cannot approve it.

    approval_answer.py, refusal
  • Command deny-list

    Known-destructive shell commands are refused whatever the mode or the approval. The list reads a command's text.

    shell_denylist.py, BUILTIN_DENIED_COMMAND_PATTERNS
  • Guarded egress

    Requests out are refused for private, loopback and link-local addresses and dial the address that was checked. By default it is not a destination allow-list.

    net/guard.py, evaluate
  • Fenced outside content

    A fetched page and other outside text reach the model marked as data, not as instructions.

    security.py, fence_untrusted
  • A token by default

    The gateway asks for a local token, and the one mode without it binds to loopback only.

    auth/modes.py, effective_bind
  • Optional sign-in

    A password, argon2id-hashed, and a second factor, for reaching the dashboard from another device. Off by default and additive: the local token keeps working, so a forgotten password cannot lock you out.

    dashboard/handlers/auth.py, the login door
  • Scoped app tokens

    An app calls the gateway with a token scoped to it, through the gateway's proxy. Its own code still runs as you, and its install review says so.

    dashboard/token_auth.py, app_session_token
  • Scanner gate

    An app is scanned in quarantine before you consent to it, and a dangerous verdict cannot be overridden into an install.

    apps/app_manager.py, install
  • Tamper-evident entries

    Each security event carries an HMAC over its own record, so an entry edited in place is caught. A deleted entry is not.

    sel.py, verify_integrity

An app is code, not a collectible.

Installing one is a staged trust decision. The package stays apart until it is scanned and you have consented to what it declares.

supply_chain.py, Verdict
  1. QuarantineThe package is staged apart, inactive, and never follows a link out of its bundle.
  2. ScanThe scanner reads the staged tree, which is the tree that would be installed, and gives a verdict.
  3. ConsentYou see what it declares, including what the platform does not enforce, before anything is enabled.
  4. EnableOnly then is it installed, with access the gateway issues. A dangerous verdict ends it at the scan.

Report a vulnerability

Core's security policy says how to report one privately and what to expect back. The threat model behind this page names each trust boundary and maps the OWASP agentic risks to the code that answers them.