omni-browser
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| OMNI_BROWSER_ALLOWED_HOSTS | No | Comma-separated list of allowed hosts to limit external navigation in the MCP, for example github.com. Subdomains are accepted; wildcard hosts are not allowed. |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| browser_openB | Navigate to an HTTP(S) page; only uses this local browser profile. |
| browser_snapshotB | Read accessibility-oriented page structure (no screenshot required). |
| browser_readC | Read visible page text for research. |
| browser_clickB | Click Playwright locator; high-impact selectors need confirm=true. |
| browser_fillC | Fill a form field via Playwright selector. |
| browser_pressC | Press a key in a focused element. |
| browser_selectC | Select an option in a standard HTML select. |
| browser_tabsB | List local browser tabs. |
| browser_switch_tabC | Switch active browser tab. |
| browser_backB | Navigate one step back. |
| browser_screenshotB | Save screenshot within local browser harness directory. No arbitrary file writes. |
| github_prepare_repositoryB | Fill new GitHub repo form without creating anything. Must be logged in manually beforehand. |
| github_create_repositoryB | Click Create repository after explicit confirmation, only on prepared GitHub new-repo form. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 13 tools
Each tool maps to a distinct browser action, and descriptions distinguish the trickier pairs reasonably well (snapshot = accessibility structure vs read = visible text). Residual overlap exists between browser_snapshot/browser_read and browser_tabs/browser_switch_tab, but the wording gives an agent enough to choose correctly.
All names use snake_case with a clear domain prefix (browser_ / github_) and verb-first construction. Minor deviation: browser_back and browser_snapshot lack a noun object, but the overall pattern is predictable.
13 tools is well within the ideal band and every tool corresponds to a concrete, non-redundant browser or GitHub-form operation. No filler tools are present.
Core browsing (open, read, interact, screenshot, back) is covered, but lifecycle gaps are notable: no forward navigation, no new-tab creation, no tab close, and no reload/wait. The GitHub portion only covers a two-step repo creation flow, which is a very narrow slice of that domain.