How do I install the FlowClaw skill?

Run npx skillmds@latest add modbender/flowclaw in your terminal (requires Node.js), paste this page's agent-chat prompt into Claude, Cursor, or any MCP-connected agent, or download the SKILL.md file and copy it into your agent's skills directory.

Is FlowClaw safe to use?

This skill has not completed SkillMD's automated safety review yet. Independent scanners report: SkillSpector: PASS, Skill Scanner: PASS. Capability flags: executes scripts. SkillMD never runs a skill's scripts for you; review the SKILL.md before installing.

Which AI agents work with FlowClaw?

This skill is tagged as working with Claude Code, Claude.ai, OpenAI Codex. SKILL.md is an open format, so most agents that read a skills directory can load it too.

Is FlowClaw free to use?

Yes. Installing skills from SkillMD is free, and the skill stays under its author's original license.

Who published FlowClaw?

modbender (@modbender) published this skill. Their other Agent Skills are listed on their SkillMD profile.

FlowClaw by modbender — Agent Skill

Name: FlowClaw
Rating: 4.2 (5 reviews)
Author: modbender

FlowClaw

Unkillable LLM load balancer for OpenClaw. Monitors usage across Anthropic, Google, OpenAI, GitHub Copilot, and Ollama. Scores accounts by urgency, rotates automatically, and fails over to local models. Your agent never gets stuck.

modbender Updated Sep 9, 2026 12 repo stars

File contents

FlowClaw — LLM Usage Monitor & Load Balancer

Track usage. Balance routing. Never waste a credit.

Unified dashboard + auto-routing for all your LLM subscriptions. Uses Earliest Deadline First scheduling — accounts resetting soonest are prioritized so unused credits aren't wasted.

Supported Providers

Provider	Auth	Data Source
Anthropic Claude Max	OAuth (unlimited accounts)	`api.anthropic.com/api/oauth/usage`
Google Gemini CLI	OAuth via OpenClaw	`cloudcode-pa.googleapis.com`
Google Antigravity	codexbar	codexbar usage API
OpenAI Codex	OAuth via OpenClaw	`chatgpt.com/backend-api/wham/usage`
GitHub Copilot	OAuth via OpenClaw	`api.github.com/copilot_internal/user`
Ollama	Local (auto-detected)	`localhost:11434/api/tags`

Commands

# 📊 Usage Monitoring
flowclaw status [--fresh] [--json]     # Full provider dashboard
flowclaw monitor [--json] [--cached]   # Clean usage report (no scoring)

# 🧠 Load Balancing
flowclaw score [--json]                # Scored ranking of all accounts
flowclaw optimize [--dry-run]          # Reorder OpenClaw model priority
flowclaw auto                          # Optimize silently (for cron jobs)

# 🛠 Utilities
flowclaw test                          # Run scoring engine unit tests
flowclaw history [N]                   # Routing decision history

Setup

Anthropic (Claude Max) — unlimited accounts

claude login
bash {baseDir}/scripts/save-account.sh
# Repeat for each account

Google Gemini CLI

openclaw models auth login --provider google-gemini-cli

Google Antigravity

openclaw models auth login --provider google-antigravity
brew install --cask steipete/tap/codexbar   # Required for usage metrics

OpenAI Codex

openclaw models auth login --provider openai-codex

GitHub Copilot

openclaw models auth login-github-copilot

Ollama

brew install ollama && ollama pull qwen3:235b
# Auto-detected — no config needed

Cron Automation

# Optimize routing every 30 minutes
*/30 * * * * bash ~/clawd/skills/flowclaw/scripts/flowclaw.sh auto

Scoring Algorithm

score = urgency(0.30) + availability(0.25) + proximity(0.15) + weekly_headroom(0.20) + tier(0.10)

Factor	Formula	Measures
Urgency	`remaining / hours_to_reset`	Credits wasting per hour
Availability	`√(remaining)`	Dampened capacity
Proximity	`1 - (hours / window)`	Deadline pressure
Weekly headroom	`(100 - weekly%) / 100`	7-day capacity reserve
Tier bonus	Free=+0.8, Sub=0, Local=-0.3	Cost preference

Perishable inventory: Both 5h session and 7d weekly windows expire. When ≤6h to weekly reset, penalty is removed — burn remaining credits. Family-aware: Only swaps within same capability class (Opus↔Opus, not Opus↔Gemini).

---name: flowclawdescription: Unkillable LLM load balancer for OpenClaw. Monitors usage across Anthropic, Google, OpenAI, GitHub Copilot, and Ollama. Scores accounts by urgency, rotates automatically, and fails over to local models. Your agent never gets stuck.--- # FlowClaw — LLM Usage Monitor & Load Balancer > *Track usage. Balance routing. Never waste a credit.* Unified dashboard + auto-routing for all your LLM subscriptions. Uses **Earliest Deadline First** scheduling — accounts resetting soonest are prioritized so unused credits aren't wasted. ## Supported Providers | Provider | Auth | Data Source | |----------|------|------------| | **Anthropic Claude Max** | OAuth (unlimited accounts) | `api.anthropic.com/api/oauth/usage` | | **Google Gemini CLI** | OAuth via OpenClaw | `cloudcode-pa.googleapis.com` | | **Google Antigravity** | codexbar | codexbar usage API | | **OpenAI Codex** | OAuth via OpenClaw | `chatgpt.com/backend-api/wham/usage` | | **GitHub Copilot** | OAuth via OpenClaw | `api.github.com/copilot_internal/user` | | **Ollama** | Local (auto-detected) | `localhost:11434/api/tags` | ## Commands ```bash # 📊 Usage Monitoring flowclaw status [--fresh] [--json] # Full provider dashboard flowclaw monitor [--json] [--cached] # Clean usage report (no scoring) # 🧠 Load Balancing flowclaw score [--json] # Scored ranking of all accounts flowclaw optimize [--dry-run] # Reorder OpenClaw model priority flowclaw auto # Optimize silently (for cron jobs) # 🛠 Utilities flowclaw test # Run scoring engine unit tests flowclaw history [N] # Routing decision history ``` ## Setup ### Anthropic (Claude Max) — unlimited accounts ```bash claude login bash {baseDir}/scripts/save-account.sh # Repeat for each account ``` ### Google Gemini CLI ```bash openclaw models auth login --provider google-gemini-cli ``` ### Google Antigravity ```bash openclaw models auth login --provider google-antigravity brew install --cask steipete/tap/codexbar # Required for usage metrics ``` ### OpenAI Codex ```bash openclaw models auth login --provider openai-codex ``` ### GitHub Copilot ```bash openclaw models auth login-github-copilot ``` ### Ollama ```bash brew install ollama && ollama pull qwen3:235b # Auto-detected — no config needed ``` ### Cron Automation ```bash # Optimize routing every 30 minutes */30 * * * * bash ~/clawd/skills/flowclaw/scripts/flowclaw.sh auto ``` ## Scoring Algorithm ``` score = urgency(0.30) + availability(0.25) + proximity(0.15) + weekly_headroom(0.20) + tier(0.10) ``` | Factor | Formula | Measures | |--------|---------|----------| | Urgency | `remaining / hours_to_reset` | Credits wasting per hour | | Availability | `√(remaining)` | Dampened capacity | | Proximity | `1 - (hours / window)` | Deadline pressure | | Weekly headroom | `(100 - weekly%) / 100` | 7-day capacity reserve | | Tier bonus | Free=+0.8, Sub=0, Local=-0.3 | Cost preference | **Perishable inventory:** Both 5h session and 7d weekly windows expire. When ≤6h to weekly reset, penalty is removed — burn remaining credits. **Family-aware:** Only swaps within same capability class (Opus↔Opus, not Opus↔Gemini).

Frequently asked questions