Skillware skills · monitoring/token_limiter
Token Limiter
Deterministic token budget gate for autonomous agent loops. After each model turn the host passes cumulative usage; the skill returns CONTINUE, WARN, or FORCE_TERMINATE so runaway tasks stop before they exhaust context or budget.
Capabilities
- Hard token ceiling
- Soft warning threshold
- Cost estimates
Problems this skill solves
People search for tasks — not always "skillware":
- how to make ai stop at token limit
- agent loop token budget
- stop runaway ai agent tokens
- limit llm spend autonomous agent
- force stop ai when context too long
- ai agent cost guardrail
- prevent infinite agent loop tokens
Install
pip install skillware then SkillLoader.load_skill("monitoring/token_limiter")
FAQ
- How to make an AI agent stop at a token limit?
- Call Skillware monitoring/token_limiter after each turn with cumulative usage; on FORCE_TERMINATE your host loop must exit — deterministic, no billing API.