Nepm orchestrator

Skill hegu-1/claude-cognition-skills/skills/nepm-orchestrator

Claude Code skills for the failure modes of long-horizon AI collaboration — continuity, judgment preservation, trap detection, faithful translation.

Install

npx -y skills add hegu-1/claude-cognition-skills --skill nepm-orchestrator

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

What its author says it does

Copied from the file, not written here

Route a user request through the NEPM skill system by deciding whether the situation primarily requires kernel interpretation, trap detection, helpful-now derivation, decision snapshot building, or founder-thought translation.

The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.

SKILL.md

3.4 KB, as published. Nobody here has run it

NEPM Orchestrator Skill

Version: 0.1

Purpose

This skill acts as the routing layer for the NEPM skill system.

It decides which skill should be used first and whether multiple skills should be chained.

The goal is not maximal workflow complexity. The goal is to choose the smallest correct structural path.

Available skills

nepm-kernel
trap-detector
helpful-now-deriver
decision-snapshot-builder
founder-thought-translator

Routing logic

Route to `nepm-kernel` when

the user input is nonlinear or layered
structural interpretation is needed first
the active thread or phase is unclear
the request contains deep movement rather than surface intent alone

Route to `trap-detector` when

movement appears active but quality may be degrading
the user is looping, collapsing, or forcing convergence
execution momentum seems to exceed clarity

Route to `helpful-now-deriver` when

many responses are possible, but only one would really help
the user needs the next foothold rather than a full expansion
minimal intervention matters more than answer volume

Route to `decision-snapshot-builder` when

the user has enough movement but no stable decision surface
a discussion is becoming dense or overloaded
execution or external communication is about to begin

Route to `founder-thought-translator` when

raw thinking must become a sendable output
the judgment exists, but the packaging does not
translation across internal and external resolution is needed

Multi-skill chains

Chain A

nepm-kernel -> helpful-now-deriver

Use when the structure must first be read, then the correct next move chosen.

Chain B

nepm-kernel -> trap-detector -> helpful-now-deriver

Use when path quality is uncertain and intervention size matters.

Chain C

nepm-kernel -> decision-snapshot-builder -> founder-thought-translator

Use when a nonlinear thought stream must become a sendable artifact.

Chain D

nepm-kernel -> trap-detector -> decision-snapshot-builder

Use when execution risk is rising and crystallization is needed before proceeding.

Output contract

Return:

selected_skill
why
secondary_skill
chain_needed
routing_note

Example:

{
  "selected_skill": "decision-snapshot-builder",
  "why": "the user has enough material but no stable decision surface",
  "secondary_skill": "founder-thought-translator",
  "chain_needed": true,
  "routing_note": "build the snapshot first, then translate outward"
}

Anti-patterns

Do not:

route everything to the kernel by habit
create long chains when one skill is enough
translate before understanding
snapshot before reading the actual structural tension
use trap detection as a default warning mechanism

Success criteria

This skill is working when:

the system chooses the right entry point quickly
unnecessary complexity is avoided
the user receives the right kind of help at the right resolution