---

name: estimate_effort description: 作業計画から必要な時間とLLMクオータ（トークン消費量）の見積もりを算出し、コストパフォーマンスを評価する

Estimate Effort

This skill analyzes a proposed implementation plan or task description to estimate the time required for completion and the approximate LLM quota consumption.

Instructions

1. Task Decomposition:

   • Break down the task into phases such as "Research", "Implementation", "Testing", and "Refactoring".
   • Assess the difficulty of each phase (Low / Medium / High).

2. Time Estimation:

   • Estimate the total time (in minutes) required for the assistant.
   • LLM-Native Benchmarks:
     • Detailed Plan Boost: If the input Markdown plan specifies method names, logic, and tests, assume implementation is 10-20x faster than human equivalent (e.g., 4 hours -> 15 mins).
     • Ambiguity Penalty: Clear requirements = 10-20 mins. Vague research tasks = 30-60 mins.
     • Test Loop: Assume pytest-driven debugging is near-instantaneous.
   • Use "Wall-clock time" (actual AI thinking/execution time) for the estimate.

3. Quota Estimation:

   • Predict token consumption based on the following criteria:
     • High: Large-scale file edits, many tool calls (run_command, replace_file_content), or long context reads. Note: High Quota != Long Time.
     • Medium: Targeted fixes, standard unit tests.
     • Low: Information retrieval only, simple renaming.
   • Account for model-specific behavior (e.g., Claude Opus is "Heavy", Gemini Flash is "Light").

4. Efficiency Evaluation:

   • Evaluate the Return on Investment (ROI) by comparing predicted results with costs (time/quota).
   • Propose "lower-cost alternative approaches" if necessary.

5. Reporting:

   • Generate an estimation report in the following format:
     • Estimated Total Time: XX minutes
     • Estimated Quota Consumption: [Low / Medium / High]
     • Breakdown: Estimates per step
     • Concerns: Uncertain elements that could cause significant time/quota overruns.