Agentic eXecution & Orchestration Systemv5.3.0
Govern AI-assisted delivery from intent to accepted outcome.
AEXOS is a CLI-first governed execution system for AI-assisted software delivery—connecting agents, task contracts, workflows, quality gates and auditability into one coherent system of record.
01 / Product truth
A task-first execution system—not a prompt library.
The product installs the team, procedures and routing layer into a project. Agents own roles and authority; tasks hold the repeatable method; workflows define transitions; checklists determine completion.
What it is
A local, CLI-first orchestration framework for AI-assisted planning and software delivery across supported agent surfaces.
How it works
The master router reads a generated squad registry, maps intent to domain ownership and routes work to the correct specialist and procedure.
Why it exists
To reduce inconsistent planning, context loss, authority drift and unverified output across AI-assisted development.
CLI is the product.
Every capability must work from the command line before it gains observability or UI.
Method lives in procedures.
Agents route and execute; the task contract carries inputs, outputs, constraints and completion gates.
Evidence closes the loop.
Story files, QA verdicts, validators and state artifacts make delivery inspectable and resumable.
02 / Operating model
Six primitives make execution legible.
AEXOS separates role, method, output shape, validation, knowledge and sequence. That separation makes the system editable, testable and portable across runtimes.
A
Agent
Role, persona, authority boundary and procedures the role may run.
T
Task
Executable procedure with declared inputs, outputs and a completion checklist.
TPL
Template
Standardized shape for PRDs, stories, architectures, reports and other artifacts.
✓
Checklist
Exit gate that determines whether the task is complete and acceptable.
KB
Data
Reference models, preferences, signal lists and domain knowledge procedures read.
WF
Workflow
Connected tasks with conditions for advancing, revising, escalating or stopping.
03 / Technical architecture
Local core, generated routing, canonical state and governed execution.
The current product is a Node.js/CommonJS system centered on .aexos-core, file-based contracts, installers, CLI entrypoints, registry generation and runtime projections for multiple AI development environments.
@aexos/core
NPM package, installer and CLI binaries.
.aexos-core
Core, Constitution, development artifacts and infrastructure.
Squad registry
Generated from squad manifests; read by the master router.
Tasks & workflows
Task-first contracts, templates, checklists and transitions.
Agents & tools
Authorized roles, external executors and local commands.
IDE projections
Claude, Codex, Gemini, Cursor, Copilot and other host formats.
Primary design rule: canonical artifacts live in the framework and squads; runtime-specific folders are derived projections, not the source of truth.
BobOrchestrator
Project-state decision tree for greenfield, brownfield and story-driven routing.
ActiveMasterOrchestrator
Coordinates the legacy/ADE Epic 0 pipeline, gates, recovery and agent invocation.
Active legacy / ADEWorkflowOrchestrator
Executes generic multi-agent phases loaded from workflow YAML.
ActiveSessionState
Canonical persistent context for new epic and story orchestration.
Canonical stateWorkflowNavigator
Read-only command suggestions from patterns and current context.
Active helperWorkflowStateManager
Compatibility layer for legacy guided workflow entrypoints only.
Deprecated compatibilityRecovery sequence documented by the architecture: validate the bound story file, load SessionState, then use the local checkpoint only as a reconciled resume hint.
04 / Workflows
Planning and delivery advance through explicit transitions.
The repository documents four primary operating paths. Each uses named owners, artifacts, verdicts and escalation rules rather than an opaque chain of agent conversations.
Draft the story from an epic or PRD with complete implementation context.
Run the 10-point story check and return GO or NO-GO with required corrections.
Implement against acceptance criteria in interactive, preflight or YOLO mode.
Issue PASS, CONCERNS, FAIL or WAIVED and attach evidence to the story.
Perform the exclusive remote push, PR and release operations after gates pass.
Inspect implementation and issue a structured verdict.
Convert findings into an addressable remediation list.
Apply QA fixes while preserving story traceability.
Repeat to approval, rejection or escalation; maximum five iterations.
Gather requirements and establish traceable functional, non-functional and constraint inputs.
Score complexity across scope, integration, infrastructure, knowledge and risk.
Research dependencies and evidence required to avoid invention.
Write, critique and convert the approved spec into an implementation plan.
Collect system structure, dependencies, behavior and operating constraints.
Identify technical debt, risks, hotspots and modernization opportunities.
Challenge the findings and reconcile evidence across specialists.
Produce executive assessment, technical report and an epic with development-ready stories.
05 / Roles & squads
Twelve core roles govern the software lifecycle.
Each role has a persona, scope, procedures and authority boundary. The latest generated registry additionally reports twenty squads and 156 projected agent definitions across software and business domains.
@aexos-masterZeusMaster orchestrator & framework developer@analystSiriusBusiness analyst@pmJanusProduct manager@poThemisProduct owner@smChronosScrum master@architectVegaSoftware architect@devVulcanFull-stack developer@qaArgusTest architect & quality advisor@data-engineerCeresDatabase architect & operations engineer@devopsPolarisRepository & release authority@ux-design-expertIrisUX/UI and design-system architect@squad-creatorArkantosSquad designer and scaffolderLatest registry: 20 squads / 156 projected agents
06 / Runtime surfaces
One roster, projected into each host’s native format.
Capabilities vary by host. Where lifecycle hooks are unavailable, the framework still operates, but checks must be triggered manually rather than firing automatically.
| Platform | Activation | Lifecycle hooks | Operational consequence |
|---|---|---|---|
| Claude Code | /agent-name | Full reference | Full automation, guardrails and audit trail. |
| Gemini CLI | /aexos-menu → agent | Native events | Minor timing differences from the reference host. |
| Codex CLI | /skills → skill | Partial | Some validators require manual trigger; leans on AGENTS.md and MCP. |
| Cursor | @agent + synced rules | None | No pre/post-action hooks; run validators manually. |
| GitHub Copilot | Chat modes + repo instructions | None | More manual workflow and validation. |
| AntiGravity | Generated workflows | Workflow-based | Uses workflow equivalents rather than native hook parity. |
07 / Installation & CLI
First value is designed to arrive in ten minutes.
The public Core package is installed from npm. New and existing project flows are intentionally different, and the framework includes doctor, validation, repair, update and uninstall paths.
Initialize a new project
init creates the named directory and installs AEXOS into it.
npx @aexos/core init my-project cd my-project
Restart the AI development environment after installation so it reloads the generated commands, skills and rules.
Add AEXOS to an existing repository
install takes no project name and writes into the current directory.
cd my-existing-project npx @aexos/core install
The installer detects an existing installation and updates in place rather than overwriting blindly.
Validate the installation
Use diagnostics, integrity validation and update commands before assuming the runtime is ready.
npx @aexos/core --versionaexos doctor --fixaexos validate --repairaexos update --checkPlan and govern delivery from the CLI
Representative execution commands exposed by the current CLI surface.
aexos sdc plan <story.md>aexos sdc next <story-id>aexos wave plan --stories a,baexos workers search <query>08 / Governance & quality
The Constitution defines non-negotiable operating constraints.
AEXOS is designed to constrain authority, require traceability and stop work when critical gates fail. Governance is embedded in procedures and validators—not left as optional advice.
CLI First
Capability exists only when it is fully operable from the command line.
Agent Authority
Exclusive authorities cannot be assumed by another role; delegation is required.
Story-Driven
Product-code work binds to a valid story with acceptance criteria and traceability.
No Invention
Specifications must trace statements to requirements, constraints or verified research.
Quality First
Lint, typecheck, tests and build gates must pass before remote publication.
Squad Portability
Squads are source of truth; runtime projections remain derived and host-agnostic.
Model Governance
Automated dispatch requires budget ceilings, routing authority, story binding and intent scanning.
Block / Warn / Info
Severity determines whether execution stops, proceeds with warning or records guidance.
09 / Ideal customers
Start where coding-agent speed is creating an accountability problem.
The strongest early buyers already use AI coding tools, repeat similar delivery workflows and feel the cost of inconsistent planning, context fragmentation and weak evidence.
AI-native software agencies
5–50 developers, repeated client delivery, margin pressure and a strong need for standardized evidence and quality.
Technical founders
Building with a small team and multiple AI surfaces, but needing a repeatable product-development operating method.
Product engineering teams
Adopting coding agents across repositories and needing roles, gates and shared delivery procedures.
Platform / AI enablement
Standardizing AI-assisted delivery for 50–500 engineers while preserving governance and portability.
Anti-ICP: buyers seeking only a prompt bundle; teams with no technical owner; organizations unwilling to define acceptance criteria; projects with no repeatable workflow or budget; anyone demanding unbounded autonomy without controls.
10 / Commercial model
Free adoption, paid acceleration and enterprise rights.
The repository establishes a broad no-fee Core grant and a separately licensed Pro edition. The packaging and price points below are a recommended GTM model, not a currently published billing schedule.
Commercial recommendation
Pricing must be validated through paid launch partners. Features marked Beta or Roadmap should not be sold as generally available until they pass explicit release gates.
MonthlyAnnual
Core Edition
$0
Personal, internal, educational and commercial use under the AEXOS license.
- CLI installer and runtime
- Core agents, tasks and workflows
- Multi-IDE projections
- Doctor, validation and update
- Community documentation
Pro Developer
$39 / month
$390 / year
Recommended individual commercial package for advanced capabilities and premium packs.
- Everything in Core
- Licensed Pro CLI access
- Premium execution packs
- Extended local history
- Email support
Team License
$499 / month
$4,990 / year
Five seats. Recommended launch package for agencies and AI-native product teams.
- Everything in Pro
- Private squad and pack rights
- Shared governance templates
- Implementation onboarding
- Priority support
Enterprise
$30K+ / year
Custom rights, enablement and support for larger engineering organizations.
- Commercial licensing terms
- Private deployment guidance
- Custom squads and policies
- Security and governance review
- SLA and premium support
Launch Partner Sprint
Four-week implementation for one priority workflow, two repositories, three gates, baseline and training.
Custom Squad / Execution Pack
Domain roles, tasks, templates, checklists, workflows and acceptance criteria packaged for reuse.
Enterprise Enablement
Architecture, governance, IDE rollout, private customization, training and operating model design.
| Access surface | Core | Pro | Team | Enterprise |
|---|---|---|---|---|
| Core CLI, roles, tasks, workflows | Included | Included | Included | Included |
| Premium squads / execution packs | Public only | Licensed catalog | Private use | Custom portfolio |
| Team governance and rollout support | — | — | Priority guidance | Custom program |
| Redistribution / embedded-product rights | Restricted | Restricted | Contract-specific | Negotiated rights |
| SLA, security review, private support | — | Priority | Contracted |
11 / Go-to-market
Land with one governed workflow. Expand through reusable packs.
The recommended motion uses the free Core to earn technical trust, then monetizes implementation, commercial rights, premium methods and enterprise support.
01
Adopt Core
Install from npm, activate one role and complete one useful command.
02
Benchmark
Choose a repeated workflow and measure baseline speed, defects and intervention.
03
Launch Sprint
Configure the workflow, gates, roles and evidence in a paid implementation.
04
License
Convert the validated workflow into Team or Enterprise commercial rights.
05
Expand
Add private packs, repositories, departments and governance coverage.
Product truth and activation
- Sanitize public naming and package references.
- Instrument npm install → first agent activation.
- Recruit 10 design partners in the primary ICP.
- Publish two source-verified technical demonstrations.
Paid proof
- Sell three Launch Partner Sprints.
- Measure accepted-output and intervention baselines.
- Package the first two premium execution packs.
- Define Pro beta release gates and support boundaries.
Repeatability
- Convert at least two partners to recurring licenses.
- Publish one quantified case study.
- Establish partner/referral motion with agencies.
- Decide which hosted control-plane capabilities are justified.
12 / Product readiness
Separate implemented capability from commercial ambition.
The public page should never imply that a cloud control plane, team workspace or enterprise identity system already exists merely because those capabilities fit the future business model.
Current product
- NPM-delivered Core CLI
- Installer, doctor, validate, update and uninstall
- Core agents, tasks, templates, checklists and workflows
- Generated squad registry and multiple IDE projections
- Story-driven workflows and local recoverable state
- Constitution, authority constraints and quality validators
Beta or partial surfaces
- Pro CLI package and license flows
- External executor delegation paths
- Graph and observability components
- Enterprise upgrade scaffolding
- Some host-specific lifecycle parity
- Newer registry content beyond README counters
Future control plane
- Hosted multi-tenant workspace
- Cloud evidence ledger and telemetry
- SSO, SCIM and enterprise RBAC
- Hosted private pack registry
- Billing and usage metering
- Centralized policy administration
13 / Product identity
One logo. One controlled identity system.
The approved AEXOS lockup appears once—in the primary header. Every other surface relies on typography, metallic execution imagery, structured spacing and teal status cues so the brand remains scarce, recognizable and premium.
Identity rule applied
Product-specific geometry. Cyryx-endorsed discipline.
The execution-stack language is expressed through spatial planes, task layers and a restrained teal path—not through repeated logo placement. This keeps the interface operational rather than promotional.
Onyx foundation#050607
Brushed structureSteel / Silver
Activation signalTeal ≤ 2%
Reserve the full AEXOS lockup for the primary header; do not repeat it in the hero, section stages or footer.
Typography, spacing, structured borders and execution-stack imagery should create recognition without logo repetition.
Use teal for status, focus and activation—not as the dominant surface or decorative glow.
14 / Frequently asked questions
What buyers and technical teams need to know.
Is AEXOS an autonomous software company?
No. The verified product is a CLI-first orchestration framework that structures AI-assisted work through roles, task contracts, workflows, authority boundaries and gates. Human review remains part of the operating model.
Is the Core Edition free for commercial work?
Yes, subject to the proprietary AEXOS license. Core may be used to build and deliver commercial work, but redistribution, forks, public mirrors and hosted offerings whose primary value is AEXOS itself are restricted.
Does AEXOS require a dashboard?
No. The Constitution explicitly establishes CLI first, observability second and UI third. A UI may improve management or visualization, but it is not required to operate the system.
Are the Pro, Team and Enterprise prices live?
No. They are recommended commercial packaging for validation with paid design partners. The repository verifies a separate Pro licensing boundary, but not a complete public SaaS billing system.
Why do README counts differ from the latest registry and QA gate?
The repository contains artifacts produced at different points in time. This page prioritizes the generated registry and latest documented QA gate while explicitly labeling those source states rather than silently combining counters.
15 / Source register
Facts are separated from commercial recommendations.
Technical claims on this page were reconciled against the public repository. Pricing, ICP sequencing and future control-plane features are identified as recommendations or roadmap targets.
README.mdProduct definition, mental model, install, roles, workflows and IDE matrix.
OPEN ↗
package.jsonPackage name, version, binaries, runtime, dependencies and validation scripts.
OPEN ↗
ConstitutionCLI First, Agent Authority, Story-Driven Development and Model Governance.
OPEN ↗
Orchestration hierarchyModule boundaries, canonical state and checkpoint reconciliation.
OPEN ↗
Generated squad registryLatest registered squads and projected agent definitions.
OPEN ↗
AEXOS Proprietary License v1.1Core grant, restrictions, ownership and separate Pro agreement.
OPEN ↗
Cyryx Labs / AEXOS
Move from AI-assisted activity to governed execution.
Start with one recurring delivery workflow. Establish a measurable baseline, configure roles and gates, and validate whether AEXOS improves accepted outcomes.