Platforms already study how you think, then use it for themselves. Monte captures your taste, judgment, and goals as a signal you own, so every agent can adapt before the work begins.
Works with your favorite agents
Claude Code
Terminal agent
Run Monte before a coding session and paste the task-aware guidance into Claude Code.
Product surface
Monte keeps the instructions your agents need close to the work, so every task starts with your preferences already in the room.
Outcome
Give every tool the same durable picture of how you work. The result is less prompt babysitting, fewer weird first attempts, and cleaner handoffs between tools.
Before
Paste the same preferences into every chat.
With Monte
The right signal arrives before the first response.
Result
Sharper plans, calmer tradeoffs, better follow-through.
Persona model
Output surfaces
monte personalize context/personalization/contextportable instruction blockSignal coverage
8
Sources
128
Memories
23
Rules
v5
Version
Prompt handoff
<MONTE_CONTEXT>
planning: reversible steps
tone: concise, evidence-first
avoid: premature abstraction
</MONTE_CONTEXT>
Developers
Install the CLI, store your Monte key, ask for task-aware guidance, then let the agent begin with the parts of you it normally has to rediscover.
Monte personalize
Real agent handoff flow
npm i -g monte-engine
monte auth
monte personalize context \
"Review this pull request before I merge it" \
--mode planning \
--jsonResponse
200 OKsurface
Primary Monte workflow
output
Task-aware instruction block
context
task planning mode
ready
Paste into the agent session
Workflow
Start with the CLI, a starter persona, or structured exports that capture how you work.
Monte keeps evidence auditable, then recalls only what matters for the task at hand.
Each task gets a compact patch: response contract, operating notes, avoidances, friction, and uncertainty.
Use the CLI, API, dashboard, or a clean instruction block before the agent does the work.
Why Monte
Your agents should start with the same durable understanding: preferences, constraints, evidence, and decision rules.
Agents start with how you decide, learn, handle risk, seek information, and respond under pressure.
Your profile can improve over time without losing what changed or why an agent used it.
Hosted keys make the same personal signal available from CLI sessions, apps, and automation flows.
Preference signals stay rule-based and reviewable, with deeper simulation workflows kept separate.
Pinned preferences, evidence, and profile metadata become portable personalization.
The dashboard gives teams a scannable way to inspect, tune, and hand off personas.
Start
Try the starter persona, then add hosted usage when your agents are saving real explanation time in real workflows.