Direct PowerShell
Run literal PowerShell commands and scripts with syntax/preflight inspection, local policy controls, streaming output, cancellation, timeouts and output limits.
POLICY-CONTROLLED COMMAND EXECUTION
PowerShell for experts.
Plain language for everyone else.
AQORIN turns operator intent into inspected PowerShell, independently assesses risk, applies local policy, and executes only within defined safety boundaries. Use exact commands when you know them. Use bounded Agent Mode when you do not. Save what worked as an Operation, let a schedule or a trigger start it, and bring your own programs as modules.
POWERSHELL BOUNDED AGENT MODE LOCAL POLICY ENGINE SAVED OPERATIONS MODULES AUDITED RUNS
CAPABILITIES
One console. Three modes. Work that runs itself.
AQORIN uses PowerShell as its first execution tool and adds local inspection, policy, bounded agent orchestration and execution controls around it - then saved Operations, schedules, triggers and modules on top. The full reference lists everything the current release supports.
Agent Mode
Describe the outcome you need. AQORIN can plan bounded PowerShell actions, evaluate each step locally, execute permitted operations, observe the result and adapt within configured limits.
Policy Engine
AI risk labels are informational only. AQORIN independently inspects PowerShell, recalculates risk locally and applies configured allow, confirm, strong-confirm or block decisions.
Command Preview
Generated PowerShell is surfaced as a structured plan with explanation, expected effect, provenance and locally calculated risk before execution controls are applied.
Operational Context
AQORIN supplies bounded machine context such as OS, shell, elevation and working directory while redacting secrets before provider context and saved history.
Operations, Schedules, Triggers
Save a goal or a reviewed command as an Operation with parameters and its own budgets. Run it by hand, on a schedule, when a file appears, a process starts, a service changes, a disk fills, load rises or a module watch reports - every run through the same inspection path. Delegate authority in advance for unattended runs, pinned to the exact revision you approved.
Modules
A folder with a module.json and the programs it names. Approve it once at exactly its content; its commands become Agent capabilities and its watches become triggers. Every module command runs as one literal, quoted PowerShell line through the same policy as a typed command. A weather module ships as the worked example.
Rendered, Reviewable Output
Answers render as full Markdown with syntax-coloured code, a raw view and Copy per block. Every card is timestamped, every Agent run is one card with its steps behind Expand, and Review Run shows the record - command, provenance, risk, policy, authorization - beside the model's story.
API Access
Create and manage scoped AQORIN credentials for future clients and integrations. Runtime API enforcement is integration infrastructure, not part of the v1.x desktop execution path.
CONTROL LOOP
Intent → plan → inspect → authorize → execute → observe.
The model proposes actions. AQORIN independently evaluates them, applies policy, and only requests human confirmation when policy requires it.
01InterpretUnderstand the operator goal and bounded machine context.
02PlanProduce a final answer, clarification or one PowerShell tool action.
03InspectRun syntax, command, parameter and local risk analysis independently of the model.
04AuthorizeApply policy and request confirmation or strong confirmation only when required.
05ExecuteRun with streaming output, cancellation, timeout and output limits.
06ObserveBound and redact the result, audit the step, then continue or stop the agent loop.
INTEGRATION ACCESS · PREVIEW
Prepare scoped credentials for AQORIN integrations.
The website can issue and manage scoped credentials for future AQORIN clients and integrations. Secrets are generated server-side and shown once. The public runtime API and scope enforcement are not yet part of the v1.x desktop product.
INTEGRATION KEY CREATION
Credential management is live. Runtime AQORIN API usage is preview infrastructure.
◎
Local policy remains authoritative
AI output is treated as a proposal, not a safety decision. AQORIN independently inspects commands, calculates risk locally, applies configurable policy and records Agent Mode runs with provenance, policy and authorization outcomes.