Api Consumption Optimizer
ReviewAudited by ClawScan on May 10, 2026.
Overview
The skill matches its quota-optimization theme, but it encourages continuous automated API calls that could drain quota or costs and it relies on an undeclared local helper.
Use this only if you intentionally want an agent to schedule API calls against your quota. Before installing or using it, add hard limits for total calls, spending, runtime, and target usage, and verify the local MiniMax status helper path yourself.
Findings (3)
Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.
An agent following this pattern could keep making API calls and use up quota or spend account budget faster than expected.
The recommended workflow performs API calls in an indefinite loop. For a provider API, this can consume quota or incur cost unless bounded by explicit user approval, a max-call count, or a budget.
while True:\n time.sleep(strategy['interval_seconds'])\n do_api_call()
Require explicit user confirmation before starting any loop, add a maximum call count or spending/quota budget, and stop automatically when the requested target usage is reached.
Users may believe the skill will preserve a quota buffer, while the provided logic can continue steering calls based on the full quota.
The documentation advertises a 90% target and 10% buffer, but the included implementation calculates strategy against total_quota and does not expose or enforce a target_usage parameter.
"target_usage": "目标消耗比例(默认0.9，即90%)" ... "建议保留10%缓冲避免突发需求"
Implement and document an enforced target_usage limit, and make the stop condition visible before any automated API-calling workflow begins.
The skill may fail or run unreviewed local code depending on what exists at that path.
The skill executes a hard-coded local helper binary to read MiniMax status, but that binary is not included in the manifest and the metadata declares no required binaries.
subprocess.run(\n ['/home/garfield/.local/bin/minimax-status-clean'],\n capture_output=True,
Declare the required status command, avoid hard-coded user-specific paths, and let users configure or approve the exact helper binary before execution.