remio General Agent
remio is your general agentic assitant, which has introduced a brand-new "Chat on the Left, Resources on the Right" dual-pane architecture. We have fundamentally separated knowledge management from AI interaction: the left pane is dedicated to Chat, while the right pane is dedicated to Resources and Artifacts. This layout eliminates interface drifting and allows you to seamlessly consult reference materials or edit AI-generated outputs while maintaining a deep conversation with the AI.
You can use remio as a unified AI workspace for multiple task types:
Quick slash commands
/pptx to create slides
/image to generate images
/xlsx for spreadsheet analysis
/deep-research for in-depth research
Session-level orchestration
Session config includes model, ability toggles, and enabled MCP server IDs
Examples
/deep-research Compare these three approaches and summarize pros and cons
/xlsx Analyze this sales sheet and provide trend highlights
aApp System (DEV / PROD)
aApps provide modular, scenario-based workflows, like a sub-agent.
Dual-environment model
DEV: development, debugging, and experimentation
PROD: stable daily usage
When switching is available
A DEV/PROD switch appears only when both versions exist for the same aApp
Where to switch
Use the DEV/PROD badge next to the aApp in the sidebar
Best practice
Validate changes in DEV first, then switch to PROD for stable usage
Relationship Between remio Agent and aApp
In remio, Agent and aApp have distinct roles and work together closely:
Agent is the orchestration layer: it understands your natural-language intent, breaks down tasks, and decides the execution path.
aApp is the capability module: it encapsulates domain actions such as retrieval, processing, write-back, and automation.
How they collaborate: you describe the goal, and the Agent invokes one or more aApps as needed to complete it.
A simple mental model:
Agent = “understanding + orchestration”
aApp = “execution + delivery”
The benefit is straightforward: you do not need to manage complex workflows manually. You only describe intent, and the system coordinates the right capabilities.
Skill System (Market + Studio)
Skills are remio’s plugin-like capability layer, supporting both usage and creation.
Skill Market (usage side)
Browse market skills
Filter by uninstalled / featured / category
Install and uninstall skills (protected built-in skills cannot be removed)
Inspect skill details and preview SKILL.md
Skill Studio (creation side)
View My Skills
Clone installed skills into a new ID (immediately usable)
Delete My Skills
Migrate misplaced skill structures to make them visible
Test publish for internal review workflows
Examples
“Open Skill Market and show only uninstalled skills”
“Clone pdf to pdf-company for an internal version”
MCP (External Tool Integration)
MCP connects remio with external tool ecosystems.
Connection methods
Quick (OAuth): connect through supported connectors
Manual
stdio: command + args
streamable-http: url + optional token
Management operations
List, install, uninstall, and enable/disable servers
Session integration
Enabled MCP servers are saved into session config and can be used during chat
Examples
“Add an HTTP MCP server with URL and token”
“Disconnect one MCP server so chat stops using that tool”
Advanced Usage / Tips
Recommended workflow
Start with general Agent capabilities → add Skill/aApp when needed → integrate MCP for external systems
DEV/PROD best practice
Use DEV for testing and iteration; keep PROD stable for daily work
Skill reuse strategy
Install a general skill first, then clone in Skill Studio for team-specific customization
MCP security hygiene
Use least-privilege tokens; if issues occur, disconnect newly added servers first
Input efficiency
Use slash commands whenever possible to reduce UI switching
Troubleshooting & FAQ
Q: Why can’t I see the DEV/PROD switch?
A: The aApp likely has only one environment version. The switch appears only when both DEV and PROD versions exist.
Q: Why can’t I uninstall some skills?
A: Those are protected built-in skills and are intentionally non-removable.
Q: Why did MCP server installation fail?
A: Check required fields (name, command, or url).For HTTP mode, verify URL reachability and token validity.For OAuth mode, make sure authorization has completed successfully.
Q: Why doesn’t my slash command trigger?
A: The command must follow strict format and include a trailing space, e.g. /pptx or /image .
Q: Why is MCP enabled but not used in responses?
A: Confirm the server is enabled and included in the current session config.If needed, start a new session and retry.