Stagent Documentation

Stagent is a local-first AI agent operations workspace. Explore the product surfaces below to see how home, execution, projects, workflows, profiles, schedules, documents, tables, chat, environment, monitoring, cost governance, and approvals fit together.

Home & Dashboard

Workspace briefing with micro-visualizations, execution lanes, and live operational signals.

Task Execution

Status-driven task flow, AI-assisted creation, bento detail view, and bulk operations.

Projects

Portfolio organization with working directories, completion donuts, and scoped execution.

Workflows

Six execution patterns, blueprint catalog with lineage, and step-level retry.

Profiles

Technical and business-function profiles across 5 providers with auto-detect routing and supervised self-improvement.

Schedules

Clock-driven and heartbeat schedules with natural language intervals, delivery channels, and cost controls.

Documents

Drag-and-drop upload, automatic text extraction, agent context injection, and output generation.

Human-in-the-Loop

Approval controls, ambient toasts, agent handoffs, budget alerts, and rich content rendering.

Monitoring

Streaming logs with click-to-navigate, auto-pause on tab switch, and event filtering.

Cost & Usage

Five-provider spend ledger with budget enforcement, project budgets, and Ollama at $0.

Agent Integration

Dual-runtime execution with Claude SDK and Codex App Server, profile-backed policies.

Chat

Tool Catalog with @ mentions, multi-provider model selection, bidirectional Slack/Telegram channels, and Quick Access.

Environment

Scan, audit, and manage Claude Code and Codex CLI environments with health scoring and artifact discovery.

Tables

Structured data tables with spreadsheet editing, charts, workflow triggers, templates, and full agent access.

Settings

Providers & Runtimes with smart routing, Ollama local models, delivery channels, permission presets, and browser tools.