skills.rest
yakeworld
yakeworldCOMMUNITY
Explore all skills by this author →
GitHub

task-router

Routes user queries to cognitive atom chains and records execution in pipeline_trace.json.

6|1|Updated May 11, 2026
Search Tags:#workflow orchestration#research automation#query classification#task routing#sub-agent delegation#pipeline trace#cognitive atoms
One-click install
npx skills add https://github.com/yakeworld/Synthos --skill task-router-yakeworld
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: task-router
Source: https://github.com/yakeworld/Synthos/tree/main/skills/core/task-router
Command: npx skills add https://github.com/yakeworld/Synthos --skill task-router-yakeworld

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? In a multi-skill research system, deciding which skills to invoke, in what order, and under which execution mode is error-prone. This Skill acts as the system entry point: it analyzes each user query, classifies its complexity, selects an execution mode (standard chain, exploratory loop, research double-loop, or parallel), and dispatches work to the correct cognitive atoms while logging every decision. ## Core Features & Use Cases - Query-to-Mode Routing: Classifies queries into standard chain, exploratory loop, research double-loop, or parallel execution, then selects the matching atom chain (ACQ→EXT→ASC→HYP→ARG→VER) without skipping or over-invoking steps. - Sub-Agent Delegation Discipline: Enforces delegate_task rules—pass the user's original goal with an empty context, never micro-manage sub-agents, and split batches of more than 10 papers into parallel subtasks. - Traceable Execution: Creates outputs/{session_id}/pipeline_trace.json recording mode, chain, per-atom status, loop state, and gene activation metadata for reproducibility. - Use Case: A user asks to "search 3D nystagmus literature"; the router selects the standard chain (knowledge-acquisition → knowledge-extraction), creates the pipeline trace, and delegates the task verbatim to a sub-agent. ## Quick Start Ask the agent to route the query "search 3D nystagmus literature" and verify that pipeline_trace.json records a standard route with the knowledge-acquisition and knowledge-extraction atoms.

Frequently Asked Questions about task-router

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I route a user query to the right skill chain?▼

Analyze the query intent and map it to an execution mode: search or extract tasks use the standard chain (ACQ→EXT), optimization questions use the exploratory loop (HYP→ARG→VER), and full research tasks use the research double-loop across all atoms. Record the decision in pipeline_trace.json.

How should delegate_task be called for sub-agents?▼

Pass the user's original words as the goal and an empty string as context. Sub-agents have their own skill library and routing rules, so micro-management instructions in context cause interruptions; tested runs show trusted delegation succeeds where micro-managed versions fail.

When should a task use parallel execution mode?▼

Use parallel mode when the query contains independent subtasks, such as searching two research directions simultaneously, or when a batch exceeds ten papers. Each subtask runs its own atom chain independently and results are merged afterward.

What is pipeline_trace.json used for?▼

pipeline_trace.json is the routing evidence record created per session under outputs/{session_id}/. It stores the session ID, selected mode, atom chain, per-atom status, loop state, and gene activation metadata so every routing decision is auditable and reproducible.

What are the limitations of the task router?▼

The router performs no cognitive operations itself; it only classifies queries, orchestrates chains, and tracks loop state. Content judgments such as whether results support a hypothesis belong to the HYP and VER atoms, and one-shot queries should not use loop modes.

Related Skills

Explore More Education & Research Skills→
allenaiallenai
30

Asta Literature Reports

Search literature, manage BibTeX citations, and update academic reports.

Official
Intermediate
lightrain-alightrain-a

paper-writing

Generate a compiled LaTeX paper PDF from a narrative report.

Community
Advanced
RowtionRowtion

cobrapy

Perform constraint-based metabolic modeling and analysis with the COBRApy Python library.

Community
Advanced
modu-aimodu-ai
1.2k

moai-foundation-cc

Author Claude Code Skills, sub-agents, plugins, slash commands, hooks, and settings.

Official
Advanced