Skills context-window-economics

Inference cost allocation and billing for autonomous AI agent collaborations. Shapley-fair cost splitting, congestion pricing, token metering, and settlement reports for context window usage. The economic layer of the Agent Trust Stack.

install

source · Clone the upstream repo

git clone https://github.com/openclaw/skills

Claude Code · Install into ~/.claude/skills/

T=$(mktemp -d) && git clone --depth=1 https://github.com/openclaw/skills "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/alexfleetcommander/context-window-economics" ~/.claude/skills/clawdbot-skills-context-window-economics && rm -rf "$T"

manifest: skills/alexfleetcommander/context-window-economics/SKILL.md

source content

Context Window Economics Protocol (CWEP)

You have an inference cost tracking and allocation system. Use it to fairly split context window costs when collaborating with other agents.

Setup

pip install context-window-economics

When to Use This Skill

When collaborating with other agents and context window costs need allocation
When tracking your own inference costs across tasks
When negotiating cost-sharing before a multi-agent transaction
When settling costs after collaborative work is complete

Core Operations

Track Context Window Usage

from context_window_economics import CostTracker

tracker = CostTracker("costs.jsonl")
tracker.log_usage(
    agent_id="your-agent-id",
    transaction_id="tx-123",
    input_tokens=2500,
    output_tokens=800,
    model="claude-sonnet-4-6",
    cost_usd=0.012
)

Shapley-Fair Cost Allocation

When multiple agents contribute to a task, allocate costs fairly:

from context_window_economics import allocate_costs

allocation = allocate_costs(
    transaction_id="tx-123",
    contributions={
        "agent-a": {"input_tokens": 3000, "output_tokens": 1200},
        "agent-b": {"input_tokens": 1500, "output_tokens": 600},
        "agent-c": {"input_tokens": 500, "output_tokens": 200}
    },
    method="shapley",
    total_cost=0.045
)
for agent_id, share in allocation.items():
    print(f"{agent_id}: ${share:.4f}")

Congestion Pricing

When context window capacity is limited:

from context_window_economics import congestion_price

price = congestion_price(
    current_utilization=0.85,  # 85% of context window used
    base_rate=0.01,
    surge_threshold=0.75,
    surge_multiplier=1.5
)
print(f"Current rate: ${price:.4f}/1K tokens")

Settlement Report

from context_window_economics import settlement_report

report = settlement_report(
    cost_file="costs.jsonl",
    period_hours=24,
    agent_id="your-agent-id"
)
print(f"Total spent: ${report.total_cost:.4f}")
print(f"Transactions: {report.transaction_count}")
print(f"Avg cost/tx: ${report.avg_cost:.4f}")

Cost Allocation Methods

Method	Description
`shapley`	Shapley value — mathematically fair based on marginal contribution
`proportional`	Split by token usage proportion
`nash`	Nash bargaining for bilateral settlement
`fixed`	Pre-agreed fixed split

Rules

Track all inference costs. Log usage for every agent-to-agent transaction.
Agree on method upfront. Cost allocation method should be in the service agreement.
Settle promptly. Generate settlement reports within 24 hours of task completion.

Security & Transparency Disclosure

Product: Context Window Economics Skill for OpenClaw Type: Skill Module Version: 0.1.0 Built by: AB Support / Vibe Agent Making Contact: alex@vibeagentmaking.com

What it accesses:

Reads and writes cost tracking files (
```
.jsonl
```
) in your working directory
No network access for core operations
No telemetry, no phone-home, no data collection

What it cannot do:

Cannot access files outside your working directory beyond what you explicitly specify
Cannot make purchases, send emails, or take irreversible actions
Cannot access credentials, environment variables, or secrets
Does not execute payments — cost allocations are recorded, not processed

License: Apache 2.0