TacoSkill LAB

The marketplace for AI agent skills

Product

  • SkillHub
  • Playground
  • Create
  • SkillKit

Resources

  • Privacy
  • Terms
  • About

Platforms

  • Claude Code
  • Cursor
  • Codex CLI
  • Gemini CLI
  • OpenCode

© 2026 TacoSkill LAB. All rights reserved.

TacoSkill LAB
TacoSkill LAB
HomeSkillHubCreatePlaygroundSkillKit
  1. Home
  2. /
  3. SkillHub
  4. /
  5. dynamic-budget-orchestrator
Improve

dynamic-budget-orchestrator

4.3

by majiayu000

88Favorites
123Upvotes
0Downvotes

Dynamically scale model token budgets using resource telemetry, prompt size, and profile presets. Use when token limits must adapt to hardware constraints, per-request size, or safe/fast/quality modes.

token-budgets

4.3

Rating

0

Installs

AI & LLM

Category

Quick Review

The skill addresses dynamic token budget allocation based on resource telemetry and profiles—a useful LLM operations capability. The description clearly states when to use it (hardware constraints, per-request sizing, profile modes). However, descriptionCoverage is moderate because a CLI agent would need more detail on input parameters and output format. taskKnowledge scores mid-range as the workflow outlines four clear steps and references a Python script and JSON presets (assumed present), but lacks specifics on how to pass parameters or interpret results. structure is reasonable with a concise overview and references to external files. novelty is modest; while helpful for orchestration, resource-aware token budgeting is conceptually straightforward and could be replicated with moderate CLI scripting, reducing the unique value proposition.

LLM Signals

Description coverage6
Task knowledge5
Structure6
Novelty4

GitHub Signals

49
7
1
1
Last commit 0 days ago

Publisher

majiayu000

majiayu000

Skill Author

Related Skills

mcp-developerprompt-engineerfine-tuning-expert

Loading SKILL.md…

Try onlineView on GitHub

Publisher

majiayu000 avatar
majiayu000

Skill Author

Related Skills

mcp-developer

Jeffallan

6.4

prompt-engineer

Jeffallan

7.0

fine-tuning-expert

Jeffallan

6.4

rag-architect

Jeffallan

7.0
Try online