think-tool

think-tool is a cursor rule for Cursor from sijeeshmiziha/visionagent. It costs 403 tokens per session, scanned A, original, MIT.

Guidance for deciding when an AI tool should pause and reason about tool results before continuing. It distinguishes this from built-in extended reasoning provided by some AI systems.

In plain words
What is it for?
Use it when designing agents that make several sequential tool calls, check policies, verify completeness, or analyze results before taking the next action.
Why use it?
It helps avoid both unnecessary reasoning steps and mistakes in long, policy-sensitive, or dependent tool workflows.

Cursor rule for Cursor

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add rules/sijeeshmiziha/visionagent/think-tool
Clone the repo
git clone --depth 1 https://github.com/sijeeshmiziha/visionagent

Made for: Cursor.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for think-tool

README.md
[![agentmods](https://agentmods.dev/badge/rules/sijeeshmiziha/visionagent/think-tool.svg)](https://agentmods.dev/rules/sijeeshmiziha/visionagent/think-tool)
Your own site
<a href="https://agentmods.dev/rules/sijeeshmiziha/visionagent/think-tool"><img src="https://agentmods.dev/badge/rules/sijeeshmiziha/visionagent/think-tool.svg" alt="Measured on agentmods" height="20"></a>
Per session 403 This file is loaded in full into every session.
When invoked 403 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.00403 $0.00403
Opus 5 $0.00201 $0.00201
Sonnet 5 $0.00081 $0.00081
Haiku 4.5 $0.00040 $0.00040

Measured 5d ago against content hash 84c77c3298a2, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

think-tool scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 5d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

.cursor/rules/think-tool.mdc · 23 lines

What it actually says

Think Tool / Reasoning During Tool Use

Extended Thinking vs Think Tool

  • Prefer extended thinking when the provider supports it and the task is simpler (e.g. planning before acting, or no need to reason over tool outputs). Use the platform’s extended-thinking feature where available.
  • Consider a "think" tool when the model must reason after receiving tool results (e.g. long chains of tool calls, policy checks, or sequential decisions where each step depends on previous outputs). The think tool is a no-side-effect tool (e.g. single thought string, handler that only logs or appends to the trace) so the model can “stop and think” between tool rounds.

When to Add a Think Tool

  • Add when: long chains of tool calls; policy-heavy or compliance-sensitive flows; sequential decisions where mistakes are costly; or the model needs to analyze tool output (e.g. verify rules, check completeness) before acting.
  • Skip when: single or parallel tool calls only; simple instruction following with few constraints; or extended thinking already covers the needed reasoning.

Implementation

  • Tool: Minimal spec (e.g. one parameter such as thought); handler has no side effects (log/trace only). Description should state that it does not change state and is for reasoning/scratchpad use.
  • Prompting: Provide domain-specific examples in the system prompt for when and how to use the tool (e.g. list applicable rules, verify information is complete, check compliance before acting). Put longer or complex guidance in the system prompt rather than only in the tool description.
  • Cost: Think tool increases output tokens but does not change external behavior unless the model uses it; low downside to including it when the scenario matches.
Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 5d ago First seen · 23 lines · 403 tokens per session scan A 84c77c3298a2

Subscribe to this mod's changes

think-tool is a cursor rule published in the GitHub repository sijeeshmiziha/visionagent (2 stars, last pushed 5mo ago), licensed MIT. It adds 403 tokens to every session, about $0.0020 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.