jeevesagent.security.permissions

Permission decisions for tool calls.

Three modes mirror the Claude Agent SDK so users don’t relearn:

• DEFAULT — allow non-destructive tools, ask on destructive

• ACCEPT_EDITS — auto-approve filesystem writes; otherwise like default

• BYPASS — allow everything (CI / sandbox use only)

Allow- and deny-lists win over modes; deny-list wins over allow-list. The decision flow:

1. Tool in deny-list → deny

2. Allow-list set and tool not in it → deny

3. Mode == BYPASS → allow

4. Mode == ACCEPT_EDITS and call is a non-destructive edit → allow

5. Tool is destructive → ask

6. Otherwise → allow

Classes

AllowAll	Trivial permission policy: every call is allowed.
Mode	Enum where members are also (and must be) strings
StandardPermissions	Mode + allow/deny-list permission policy.

Module Contents

class jeevesagent.security.permissions.AllowAll

Trivial permission policy: every call is allowed.

The default for Agent when no permissions are configured.

async check(call: jeevesagent.core.types.ToolCall, *, context: collections.abc.Mapping[str, Any]) → jeevesagent.core.types.PermissionDecision

class jeevesagent.security.permissions.Mode

Bases: enum.StrEnum

Enum where members are also (and must be) strings

Initialize self. See help(type(self)) for accurate signature.

ACCEPT_EDITS = 'acceptEdits'

BYPASS = 'bypassPermissions'

DEFAULT = 'default'

class jeevesagent.security.permissions.StandardPermissions(*, mode: Mode = Mode.DEFAULT, allowed_tools: list[str] | None = None, denied_tools: list[str] | None = None)

Mode + allow/deny-list permission policy.

async check(call: jeevesagent.core.types.ToolCall, *, context: collections.abc.Mapping[str, Any]) → jeevesagent.core.types.PermissionDecision

classmethod strict() → StandardPermissions

Convenience: default-mode permissions with no overrides.