GrowthClaw

Marketing execution workflows for OpenClaw, with eval gates and a task pipeline.

What You Get

• Deterministic workflows (YAML) for intake, execution, and strategy evolution
• Marketing agent roles (context, planner, executor, evaluator, PM)
• Persistent task pipeline in SQLite
• Strategy evolution cron loop (3x/day)
• Local dashboard for approvals, pipeline operations, and run visibility

Requirements

• Node.js >=20 (enforced in package.json)
• OpenClaw installed/running for full agent runtime integration
• Local shell with crontab available (for system cron fallback)

GrowthClaw uses OpenClaw cron integration when available; otherwise it installs system cron jobs with schedule + payload bindings that call GrowthClaw workflows.

Install

Installer script (after hosting this repo):

curl -fsSL https://raw.githubusercontent.com/<org>/growthclaw/main/scripts/install.sh | bash

Alternative from OpenClaw prompt:

tell OpenClaw agent: install github.com/<org>/growthclaw

Local clone install:

npm install
npm run build
npm link
growthclaw install --with-dispatcher-cron

Quickstart

growthclaw install
growthclaw workflow run intake-and-plan "Set up GrowthClaw for <company>"
growthclaw tasks list --status needs-approval
growthclaw tasks approve --all
growthclaw dashboard

Then open http://127.0.0.1:3333.

How It Works

Loop:

1. Context intake from website + answers
2. Initial strategy creation
3. Task backlog generation
4. Human approval gate
5. Task execution
6. Per-task strict eval gate
7. PM transition decision

strategy-evolution runs 3x/day and proposes only justified additions.

Task Pipeline Statuses

• backlog: created, not yet ready for approval
• needs-approval: ready for human review
• approved: execution allowed
• running: executor currently producing artifacts
• eval: awaiting evaluator verdict
• revise: evaluator requested specific changes
• blocked: human input/external blocker/escalation required
• done: complete and accepted
• archived: optional parked/retired state

Workflows Shipped

intake-and-plan

• Inputs: setup prompt / website context
• Writes: company_context, strategy_versions, tasks, task_events
• Outputs: APPROVAL_REQUEST.md

execute-task

• Inputs: approved task id
• Writes: task artifacts in ~/.growthclaw/tasks/<task-id>/, task status transitions, eval fields, task events
• Outputs: evaluator verdict and transition to done / revise / blocked

strategy-evolution

• Inputs: latest context + strategy + tasks + outcomes
• Writes: new proposal tasks (when needed), approval summary updates
• Outputs: updated APPROVAL_REQUEST.md

Dashboard

Start/stop/status commands:

growthclaw dashboard
growthclaw dashboard stop
growthclaw dashboard status

Key screens:

• Tasks Kanban
• Task detail (artifacts, eval, PM reasons, history)
• Approvals
• Strategy
• Runs
• Cron
• Logs
• Settings

All UI actions call guarded GrowthClaw backend endpoints using the same PM transition rules as CLI.

Cron Schedule

Default strategy evolution schedule (local time):

• 09:00
• 13:00
• 17:00

Manual trigger:

growthclaw workflow run strategy-evolution

Reinstall cron from dashboard or CLI install command.

Configuration

Config file:

• ~/.growthclaw/config.json

Supported keys:

• dashboard.port
• safeMode.draftOnly
• safeMode.allowForceMoves
• limits.maxRetries
• cron.strategyEvolution.times
• integrations.* placeholders

You can also relocate runtime root with GROWTHCLAW_HOME in future versions (planned).

Security Model

• Draft-only publishing defaults enabled
• Human approval gate before execution
• Retry caps for revise loops
• PM guarded transitions (no bypass unless force is explicitly allowed)
• Secrets are never committed to repo; use environment variables
• Dashboard sanitizes logs/artifacts to avoid secret leakage
• Optional dashboard basic auth via GROWTHCLAW_DASHBOARD_BASIC_AUTH=1

Build Your Own Workflows

Minimal workflow example:

id: custom-workflow
name: Custom Workflow
trigger: manual
runtime: openclaw
steps:
  - id: plan
    agent: strategy_planner
    tools: [files]
  - id: execute
    agent: executor
    tools: [files, web]
  - id: evaluate
    agent: task_eval
    tools: [files]

Keep eval/verdict gates explicit and route transitions through PM rules.

Commands

growthclaw install
growthclaw workflow list
growthclaw workflow install <id>
growthclaw workflow run <id> "<task>"
growthclaw tasks list [--status ...]
growthclaw tasks view <task-id>
growthclaw tasks approve <task-id>
growthclaw tasks approve --all
growthclaw tasks edit <task-id> [--patch '{"priority":"P0"}']
growthclaw tasks move <task-id> <status> [--force]
growthclaw tasks comment <task-id> "<text>"
growthclaw strategy current
growthclaw strategy history
growthclaw logs [<lines>]
growthclaw dashboard
growthclaw dashboard stop
growthclaw dashboard status

Contributing

• Add workflows in resources/workflows/*.yml
• Add/adjust agent rubrics in resources/agents/*.md
• Extend API in src/dashboard/server.ts
• Extend UI in dashboard/src/pages/*
• Run checks:

npm run typecheck
npm run build