Assign model tiers to subtasks using a weighted 5-factor risk matrix.
npx skills add https://github.com/hjemmesidekongen/ai --skill tier-assignment
Please help me install this Agent Skill. Skill: tier-assignment Source: https://github.com/hjemmesidekongen/ai/tree/main/plugins/smedjen/skills/tier-assignment Command: npx skills add https://github.com/hjemmesidekongen/ai --skill tier-assignment
This Skill automates the process of assigning the appropriate AI model tier (junior, senior, principal) to decomposed subtasks, ensuring cost-efficiency and optimal performance by evaluating task complexity and risk.
Use the tier-assignment skill to evaluate the risk and assign a model tier for the subtask described in 'subtask-details.yml'.
High-intent search queries and answers about installing and using this skill.
You can assign AI model tiers by evaluating decomposed subtasks against a weighted 5-factor risk matrix that analyzes scope, reversibility, ambiguity, impact, and dependencies to output junior, senior, or principal model allocations.
AI model tiering is the process of routing subtasks to junior, senior, or principal models based on risk assessment. You need it when dispatching agents to prevent overspending on simple tasks while ensuring complex tasks get sufficient capability.
Risk assessment works by scoring subtasks across five weighted factors: scope, reversibility, ambiguity, impact, and dependencies. These scores generate per-subtask justifications that determine whether a junior, senior, or principal model is required.
Yes, by applying automated task decomposition and evaluating the resulting subtasks with a risk matrix, you can assign lower-cost junior models to simple tasks and reserve expensive principal models strictly for high-impact, ambiguous, or complex operations.
To determine if a subtask needs a senior or principal AI model, the system evaluates five risk factors: task scope, reversibility, ambiguity, business impact, and dependencies, producing a justification for each tier assignment.
The risk matrix approach requires fully decomposed subtasks with clear scope to function accurately. If task decomposition is incomplete or ambiguity cannot be reliably scored, the model tier justifications and subsequent agent dispatch decisions may be suboptimal.
tkstangAdd a backlog item with YAML metadata and a generated unique ID.
ChegginTrack token usage and enforce per-task budgets across agent sessions.
HillBombCreationsAnalyze Vivreal's cost structure, pricing strategy, and scaling economics.
tiancaiamaoClassify agent tasks and route them to optimal LLM models.
