How LoopFlowHQ Works

LoopFlowHQ is a workflow designed for humans and agents to work together with clear ownership boundaries. The role badges below show who can take each phase in Next Best Action selection.

Mission: LoopFlowHQ centralise et encadre de façon traçable votre stratégie et votre implémentation en un seul flux guidé, en utilisant l’IA de votre choix à chaque étape et dans chaque contexte.

Vision: LoopFlowHQ se veut un workflow IA-native, conçu pour les product builders et solopreneurs, intégrant de bout en bout idéation, stratégie et exécution, sans copier les lourdeurs des grandes équipes, avec des agents IA partenaires.

Epic Workflow

Discovery -> Refine -> Plan -> Implement -> Close

Epic · Discovery

HumanChat AgentCoding Agent

Purpose: Understand the problem space and decide if the Epic is worth pursuing.

States: discovery:ready, discovery:in_progress, discovery:done

Start: Human or Chat Agent starts discovery:ready -> discovery:in_progress.

End: Human confirms discovery:in_progress -> discovery:done.

  • Agent can ask questions, research context, challenge assumptions, and propose kill with justification.
  • Human owns final decision to continue, rollback, or kill.

Epic · Refine

HumanChat AgentCoding Agent

Purpose: Define what success looks like before committing to execution structure.

States: refine:ready, refine:in_progress, refine:done

Start: Human or Chat Agent starts refine:ready -> refine:in_progress.

End: Human confirms refine:in_progress -> refine:done.

  • Refine output includes acceptance criteria, hypotheses, and non-goals.
  • No Story creation happens in Refine.

Epic · Plan

HumanChat AgentCoding Agent

Purpose: Translate Epic acceptance criteria into executable Stories with semantic coverage.

States: plan:ready, plan:in_progress, plan:done

Start: Coding Agent starts plan:ready -> plan:in_progress.

End: Coding Agent confirms plan:in_progress -> plan:done.

  • Agents can propose splits/merges and surface scope gaps.
  • When plan:done is reached, Story creation and Epic transition side effects can be automatic.

Epic · Implement

HumanChat AgentCoding Agent

Purpose: Track execution progress through Story aggregation, not Epic-level execution.

States: implement:ready, implement:in_progress, implement:done

Start: System-managed transition based on Story implementation progress.

End: System-managed transition when Story execution is complete.

  • Epic implement is not directly assigned via Next Best Action.

Epic · Close

HumanChat AgentCoding Agent

Purpose: Validate intent fulfillment and archive the Epic with explicit acceptance review.

States: close:ready, close:in_progress, close:done

Start: Coding Agent starts close:ready -> close:in_progress.

End: Coding Agent confirms close:in_progress -> close:done.

  • Each acceptance criterion must be evaluated (met, partial, cancelled).
  • Followup decision is mandatory (create or skip).

Story Workflow

Plan -> Implement -> Review -> Close

Story · Plan

HumanChat AgentCoding Agent

Purpose: Clarify story intent and review expectations before implementation starts.

States: plan:ready, plan:in_progress, plan:done

Start: Human or Chat Agent starts plan:ready -> plan:in_progress.

End: Human confirms plan:in_progress -> plan:done.

Story · Implement

HumanChat AgentCoding Agent

Purpose: Execute code and delivery artifacts end-to-end.

States: implement:ready, implement:in_progress, implement:done

Start: Coding Agent starts implement:ready -> implement:in_progress.

End: Coding Agent marks implement:in_progress -> implement:done.

  • Typical outputs: code changes, tests, lint/build success, PR, preview deployment.
  • PRs are not merged in Story Implement; wait for human approval in Story Review.

Story · Review

HumanChat AgentCoding Agent

Purpose: Human product validation gate.

States: review:ready, review:in_progress, review:done, review:changes_requested

Start: Human starts review:ready -> review:in_progress.

End: Human decides review:done (approved) or review:changes_requested (rework).

  • Review cannot be skipped for Story tickets.
  • Merge the Story PR only after review approval.
  • Changes requested loops the Story back to implementation readiness.

Story · Close

HumanChat AgentCoding Agent

Purpose: Explicitly mark accepted Story as complete.

States: close:ready, close:in_progress, close:done

Start: Coding Agent starts close:ready -> close:in_progress.

End: Coding Agent confirms close:in_progress -> close:done.

Task Workflow

Define -> Implement -> Close

Task · Plan (Define)

HumanChat AgentCoding Agent

Purpose: Frame and prepare the work unit before implementation.

States: define:ready, define:in_progress, define:done

Start: Chat Agent or Coding Agent starts define:ready -> define:in_progress.

End: Chat Agent or Coding Agent confirms define:in_progress -> define:done.

Task · Implement

HumanChat AgentCoding Agent

Purpose: Execute and validate the technical work.

States: implement:ready, implement:in_progress, implement:done

Start: Coding Agent starts implement:ready -> implement:in_progress.

End: Coding Agent confirms implement:in_progress -> implement:done.

Task · Close

HumanChat AgentCoding Agent

Purpose: Finalize delivery and mark the Task complete.

States: close:ready, close:in_progress, close:done

Start: Coding Agent starts close:ready -> close:in_progress.

End: Coding Agent confirms close:in_progress -> close:done.

Human Control Points

  • Start and finalize human-gated transitions in Epic and Story flows.
  • Approve or reject Story review outcomes.
  • Evaluate Epic acceptance criteria at Close.
  • Choose whether Followup should be created at Epic close.