Estimate issue complexity and output structured JSON with split suggestions.
npx skills add https://github.com/pcortes/swarm-attack --skill complexity-gate
Please help me install this Agent Skill. Skill: complexity-gate Source: https://github.com/pcortes/swarm-attack/tree/main/swarm_attack/skills/complexity-gate Command: npx skills add https://github.com/pcortes/swarm-attack --skill complexity-gate
Estimate issue complexity before implementation to prevent timeouts and guide splitting decisions.
Provide issue title, body, and acceptance criteria; returns an estimation including estimated_turns, complexity_score, needs_split, and suggestions.
High-intent search queries and answers about installing and using this skill.
Complexity estimation determines the number of LLM turns required for implementation by analyzing issue title, body, and acceptance criteria. The Skill outputs a complexity score, estimated turns, and a needs_split flag to guide whether work should be split into smaller tasks before development begins.
Split recommendations are generated when estimated LLM turns exceed complexity thresholds that risk timeouts. The Skill provides concrete split suggestions based on acceptance criteria, helping you break work into manageable chunks that fit within LLM turn budgets.
Provide the issue title, full description or body, and acceptance criteria. The Skill analyzes these inputs against complexity guidelines to calculate estimated turns, confidence levels, and detailed reasoning for its recommendations.
Yes. By estimating turns upfront and flagging issues that need splitting, you avoid starting work that exceeds LLM capacity limits. This prevents mid-implementation failures and guides better planning decisions for LLM-based development workflows.
The complexity score quantifies implementation difficulty relative to LLM turn capacity. Combined with the estimated_turns value and confidence level, it indicates whether an issue is ready to implement or requires splitting into lower-complexity subtasks.
dokipenCreate tickets in GitHub, Jira, or compatible trackers from title, description, and acceptance criteria.
TruCol270Enforce daily LLM budget hard stops before every agent call.
zhengxuyuGuide product managers through breaking epics into INVEST-compliant user stories.
blakeoxAnalyze product feature proposals for necessity and value.
