Intent drifts
The goal gets reinterpreted across chats until the Agent is solving a different problem.
A local-first working agreement for AI Agents
Give every Agent a clear way to work with you.
SoloMap turns your intent into a shared working agreement: the outcome, boundaries, authority, evidence, and project memory every Agent needs to move work forward.
Change agents or start a new session without losing what matters.
You keep final judgmentWorks with your local Agent CLIEvidence before done
Open source. Local-first. Built for VS Code.
Working agreement
01Outcome
02Boundary
03Action
04Evidence
Agreement ready
Outcome A new user can start a first project in five minutes
Boundary Setup, registration, and first run only
Authority May change onboarding; user approves before release
Local Agent run
Context Outcome, done criteria, relevant files
Action Changes and output remain visible
State Awaiting verification before completion
Evidence and continuity
Evidence Real flow ran and the result is reviewable
Decision User accepts the current result
Memory Next run continues from verified state
A prompt starts the work. It does not define the relationship.
Without a shared agreement, every new Agent session has to guess what matters, what it may change, and when the work is actually done.
Authority stays implicit
The Agent has to infer what it may change, what must stay untouched, and when to stop.
Done becomes a claim
A confident reply can look complete even when the real result has not been checked.
Four actions make Agent work reliable across sessions.
The agreement stays simple enough to use on a real task, but strong enough to survive a model or session change.
Define the outcome
Say what must become true and how you will recognize a useful result.
Set boundaries and authority
Make clear what the Agent may change, what it must protect, and where your approval is required.
Let your Agent act
Run the local Agent CLI you already use with the right project context attached.
Accept evidence and continue
Review the real result, decide whether it is done, and leave the next session a verified starting point.
SoloMap is the working agreement around the Agent.
It does not replace your coding Agent or make product decisions for you. It keeps your outcome, boundaries, authority, evidence, and project memory together so each Agent can act from the same verified state.
You decide
the outcome, boundaries, approval points, and whether the evidence is good enough.
The Agent acts
inside that agreement using the coding tool you already trust.
Evidence moves the project
so the next session starts from what was verified, not from what was merely claimed.
| Decision | SoloMap working agreement | Agent chat alone |
|---|---|---|
| Center of control | Your outcome and final judgment | The current conversation |
| Outcome and boundary | Explicit and reusable across runs | Restated or inferred in each session |
| Agent authority | Clear before the run begins | Often discovered while the Agent is acting |
| Completion | Accepted from reviewable evidence | Usually reported in the Agent reply |
| Continuity | Resumes from verified project memory | Depends on available chat context |
| Agent choice | Bring the local Agent CLI you prefer | Bound to the tool hosting the conversation |
Where the working agreement stands today.
SoloMap is moving from reliable local execution toward a portable agreement that can survive changes of Agent, session, and project.
-
Achieved
Local project truth
Roadmaps, next actions, and project memory stay beside the code in your local workspace.
-
Achieved
Verifiable Agent execution
Start your chosen local Agent from a roadmap step, then keep completion criteria, run history, handoff, and evidence.
-
Current stage
One agreement across sessions
We are bringing outcome, boundaries, authority, evidence, and memory into one consistent interaction across Agent runs.
-
Next
Agreement portability across Agents
Carry the same verified agreement into the Agent CLI you choose without rebuilding project context each time.
Final vision
A durable protocol for human-Agent work.
Whichever Agent, project, or session you use, define the outcome and boundaries once; the Agent acts within them, delivers evidence, and the project continues from verified state.
Your working agreement stays with your project.
SoloMap keeps the core project truth in your local workspace. The website does not become a cloud copy of your roadmap, prompts, or Agent history.
Outcomes, evidence, and project memory stay with the local project.
You choose which local Agent CLI performs the work.
External signals are connected or shared only when you decide.
Free establishes one project agreement. Pro helps you choose across projects.
Strategy cockpitPortfolio tradeoff viewPortfolio healthAbility compoundingMarket and delivery diagnosisEarly access roadmap input
Start with the problem you have now
Set up SoloMap, build an AI coding roadmap, or resume a project after a break.
SoloMap Quick Start
Install SoloMap, register a local project, create your first roadmap, and run a focused Agent task from VS Code.
How to Build an AI Coding Project Roadmap
Create an AI coding roadmap that keeps product goals, Agent tasks, verification, launch work, and feedback connected.
How to Resume an AI Coding Project Without Losing Context
Resume an AI-assisted coding project after a break using roadmap state, local history, touched files, and verification evidence.
OPC Blog
Turn a solo product into a durable business.
Field guides on AI collaboration, shipping, organic growth, and running a one-person company.
Project Roadmaps
AI Chat, Roadmap Step, or Autonomous Flow: Which Mode Should You Use?
Choose open conversation, bounded roadmap execution, or evidence-driven autonomous flow based on how clear the outcome is and how much safe automation the task supports.
Agent Workflows
AI Coding Agent Handoff Checklist for Codex, Claude Code, and Cursor
Create a tool-neutral handoff that lets the next coding agent continue from verified state without replaying chats or repeating failed work.
Solo Developer Systems
How to Manage Multiple Side Projects Without Losing the Week
Run a solo project portfolio with one weekly focus, clear product roles, evidence tracking, and stop rules that prevent low-leverage work from consuming the week.
SoloMap FAQ
Is SoloMap an AI coding Agent?
No. SoloMap is the local-first working agreement around the coding Agents you already use.
What is a human-Agent working agreement?
It is a shared definition of the outcome, boundaries, Agent authority, completion evidence, and project memory needed to work reliably across sessions.
Does SoloMap replace prompts or task lists?
No. Prompts can start a run and task lists can track work. SoloMap connects them to durable project intent, evidence, and a verified next state.
Who decides when work is done?
You do. SoloMap keeps your completion criteria and the Agent's evidence together so the decision is not based on the Agent's claim alone.
Can SoloMap work with different Agent CLIs?
Yes. You bring the local Agent CLI you prefer, while SoloMap preserves the working agreement around it.
Does SoloMap require a hosted backend?
No. The core workflow keeps roadmap state, task records, evidence, and project memory in the local workspace by default.
Start with one real Agent task.
Choose a project, define what done means, and let SoloMap carry that agreement into the next Agent run.