Features

Everything in one workspace.
Nothing you have to babysit.

A precise tour of what Pogama does today. If it's on this page, it's in the app.

AI

Streaming answers, reasoning, and modes for what you're doing: Chat, Ask, Plan, Build, Research, Agent.

Any model: Anthropic, OpenAI, Google Gemini, OpenRouter, Ollama, LM Studio, Any OpenAI-compatible server. Local servers are detected automatically.
Auto & fallback: “Auto” picks a model by task; if one fails, a fallback model continues.
Honest costs: Cost is shown only when the provider reports it — otherwise it says Unknown.
Use case: Ask about a bug with @file mentions; switch to a local model for private notes.

Reusable assistants with their own instructions — per project, mentionable with @.

Project teams: Give a project its own agents; one can answer every new chat.
Several at once: Mention more than one and each answers in the same turn.
Skills & memory: Activatable knowledge and a memory you can see and delete.
Use case: A “reviewer” agent that checks every change against your conventions.

Explorer, search and a Monaco editor — with every AI change reviewed before it's written.

Reviewed writes: “Apply to file” opens a side-by-side diff; apply with permission, undo any time.
Ask about a file: Send the open file to the chat with one click.
Use case: Let the AI draft a refactor, then accept it hunk by hunk.

Every repository of your projects in one place. Every action shows the exact Git command it runs.

Changes: Staged, unstaged, untracked and conflicts; stage files, hunks or single lines.
History & branches: Graph, commit details, create/switch/rename/delete branches, publish.
Safe sync: Fetch, fast-forward-only pull by default, push with an explanation.
Use case: Commit with Conventional Commits and copy the command to your own terminal.

Folders → projects → chats. A project can live only in Pogama or link a folder on disk.

Your team's work: projects, statuses, priorities, assignees, due dates — next to the code.

Real shells (PowerShell, Bash) in tabs, and `pogama` from any terminal.

Run from chat: “Run in terminal” on any shell block — always with permission.
pogama .: Open the current folder as a project, or `pogama "question"` to start a chat.
Use case: Run tests the AI suggested, after reading the exact command.

The AI can browse with Firefox (WebDriver BiDi) when you allow it.

Your cookies, if you choose: Import cookies from Firefox for sites you're signed in to — only when you start it.
Use case: Research a library's docs and summarize them into Knowledge.

Security Lab scans a project before you trust it — on your computer, offline, never executing what it reads.

Engines: Leaked secrets, install scripts, risky CI, VS Code tasks, Git hooks, obfuscated code, suspicious executables.
Honest verdicts: Caution, Review before running, or Don't run — never “safe”. Reports in Markdown, JSON or SARIF.
Use case: Scan a repository a client sent before running npm install.

Workflows, tools and MCP servers — each with its own permissions.

Workflows: Multi-step recipes that run in order, with progress and a stop button.
Tools & MCP: Tools with schemas and permissions; real MCP servers over stdio, from templates.
Use case: A release checklist that runs your scripts and drafts the changelog.