Model and agent configuration
Configure model providers, agent instructions and specialist subagents. Subagents can have their own models and allowed tool lists.
ALTAI TECHNOLOGY / isanagent
An agent harness shaped around your work.
isanagent is a harness for running and customizing agents that use tools. ALTAI configures models, instructions, tools and permissions to build research, analysis and operational workflows for your organization.
Discuss the work your agent should doCustomizable agent harnessSource code · GitHub ISANAGENTEXAMPLE CONFIGURATION
TASK
Checking & reporting agent
Gather records, flag gaps and prepare a report.
Model + instructionsisanagentTools + execution state
- Tools
- Internal API · File reading
- Skill
- Record checks · Reporting
- Permissions
- Read access; approval for changes
- Output
- Report draft with source references
A configuration illustration. Tool connections and tasks are developed for each project; no agent is running here.
01 / THE TECHNOLOGYISANAGENT
The agent harness ALTAI uses to build client-specific agents: it runs tasks, calls tools and tracks the state of the work.
Skills, plugins and tool connections
Define procedures as skills. Plugins bring together instructions, MCP tool connections, subagents and hooks.
Execution boundaries and controls
Configure tool permissions, working directories and command approval rules. Hooks can add checks before tool calls and logging after them.
Multi-step task execution
The agent uses tool results to inform its next steps. Depending on configuration, it can delegate to subagents and run code locally, in Jupyter or over SSH.
02 / WHERE IT FITSTEAMS & TASKS
A place in
your team’s work.
Examples to scope around your data, operating requirements, and the people reviewing the result.
Operations teams
Bring data collection, checking and reporting into an agent workflow connected to internal tools.
Define the steps, tool access and human review points for recurring work together.
Research and engineering
Run source research, code execution and result review within one workflow.
Produce work that can be inspected through tool outputs and execution records.
Software and data teams
Add organization-specific instructions and system connections through skills and plugins.
Create reusable agent capabilities with a defined scope for similar tasks.
03 / BUILT AROUND YOUR ENVIRONMENT
Make it work
with your systems.
ALTAI builds agent roles, internal system connections, skills, plugins, interfaces and review steps around your work. When Yada is used for model calls, routing and data policy are configured separately. Tool access and its data flows are also addressed within the solution design.
Explore custom developmentWHAT WE EVALUATE
01Correct completion and need for human intervention on representative tasks
02Tool use within defined permission and approval rules
03Time, model usage and output quality per task
Criteria are agreed for the intended task and environment.WORK WITH ALTAI
Discuss the work your agent should do
Tell us about the task, the data, and the systems involved. We’ll work out the right scope with your team.
Discuss the work your agent should do