👤 Human
Strategy & Approvals

↓

🦞 Wave — Orchestrator
Senior Technical Orchestrator

↓

Specialized Agent Team

↓

Outputs

📝 Blog Posts
🐦 X Posts
🔒 Security Audits
🎨 Social Visuals
💻 Code
📈 Daily Reports

🦞 Wave — Orchestrator

The central intelligence. Manages all sub-agents, routes tasks, maintains strategic context, and ensures every action aligns to the mission.

💻 Coder

Writes, tests, and deploys code. Builds integrations, hardens security, and automates workflows. All terminal commands require explicit human approval.

🔒 Security QA

Runs full security test suites before every deployment. Penetration testing, vulnerability audits, input validation checks. Zero-trust is the default.

🎨 Image Creator

Generates branded visuals locally — quote cards, infographics, comparison charts. Runs without external API calls. Fast turnaround, full brand control.

✍️ Content Writer

Drafts X posts, blog copy, and reply candidates in Wave’s voice. High volume, high speed. All content approved by human before publication.

🔍 Researcher

Surfaces X engagement opportunities, follow candidates, trending topics, and market intelligence. Finds signal in noise so the operator doesn’t have to.

📤 Publisher

Executes approved posts to X, publishes to WordPress, and deploys to Vercel. Pure mechanical execution — nothing ships without explicit human approval.

SOUL.md

Defines Wave’s identity, decision-making philosophy, zero-trust directives, and communication style. The personality layer.

MEMORY.md

Long-term curated memory. Key decisions, lessons learned, active projects, and context that survives across sessions.

HEARTBEAT.md

The daily operations brief. Read every morning at 8am CT by a scheduled cron job. Drives the daily status report.

project_signal.md

Single source of truth for the business launch. Tracks all ideas, blockers, assets, next actions, and signal thresholds.

x_rules.md

Complete rules of engagement for @TheWaveAgent on X. Draft-only mode, posting schedule, engagement rules, crisis protocol.

memory/YYYY-MM-DD.md

Daily session logs. Raw notes of what happened, decisions made, and what’s pending. Auto-created each day.

Zero-Trust by DefaultSecurity is the baseline. Every terminal command requires explicit human approval via Telegram.

Human in the LoopThe human is the air-gap. Wave manages and recommends — consequential decisions always route to the human.

Bias for ActionIdeas without execution are worthless. Every conversation drives toward a tangible, measurable outcome.

Do No HarmWave will never build or execute anything with potential to harm people. An absolute boundary.

Memory = ContinuityEvery significant decision is written to a file. Mental notes don’t survive session restarts. Files do.

Validate Before TrustEvery new skill passes through the AST security scanner before installation. No exceptions.