All skills
simota avatar

/nexus

@c805268
by shingo imotasimota/agent-skills85 stars
15

Orchestrating multi-specialist task chains and scope-adaptive product delivery: classifies intent, selects and executes the minimum viable chain, aggregates results, and verifies acceptance criteria. For multi-domain tasks, build-first delivery, and product lifecycle execution.

Use this Skill: https://skilld.dev/gh/simota/agent-skills/nexus

This session only. Nothing lands on disk.

referencequality-iteration.md

≈713 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Quality Iteration (absorbed from Hone)

Purpose: PDCA loop for iterative quality improvement after initial delivery. Read when: Judge feedback or user requests require iterative polishing.

PDCA-based quality improvement methodology. Previously a standalone agent (Hone), now integrated as a Nexus orchestration pattern.


When to Apply

Use this pattern when Nexus needs to iteratively improve output quality:

  • After initial implementation, quality gate fails
  • Judge reports quality issues requiring multiple specialist fixes
  • User requests "polish" or "improve quality" of existing output

PDCA Cycle

PLAN(Diagnose) → DO(Execute) → CHECK(Measure) → ACT(Learn) → repeat or terminate

PLAN: Diagnose quality gaps

  • Run Judge for code review findings
  • Measure current baseline (see UQS below)
  • Identify highest-impact improvement areas
  • Select specialist agents for fixes

DO: Execute improvements

  • Route to specialist agents in order (see coordination below)
  • Each agent addresses specific quality gap

CHECK: Measure improvement

  • Re-run Judge / relevant quality checks
  • Calculate new UQS score
  • Compute delta from previous cycle

ACT: Learn and decide

  • If target met → terminate
  • If diminishing returns → terminate
  • Otherwise → next cycle with remaining gaps

Unified Quality Score (UQS)

Agent Normalization Weight
Judge 100 - (CRIT×25 + HIGH×15 + MED×5 + LOW×2) 0.37
Consistency 100 - (HIGH×15 + MED×5 + LOW×2) 0.10
Test Quality isolation×0.25 + flaky×0.25 + edge×0.20 + mock×0.15 + read×0.15 0.10
Zen max(0, 100 - (avgCC - 10) × 5) 0.15
Radar coverage% 0.20
Quill pass_rate × 100 0.08

UQS = Σ (normalized_score_i × weight_i)

Interpretation: 90-100 Excellent · 80-89 Good · 70-79 Acceptable · 60-69 Fair · <60 Poor


Agent Coordination Order

Judge(detect) → Builder(fix critical) → Sentinel(if security) → Zen(simplify) → Radar(add tests) → Quill(document)
Quality Gap Primary Agent Skip If
Bugs detected Judge → Builder No bugs
High complexity Zen avgCC < 10
Low coverage Radar coverage > 80%
Poor docs Quill docs complete
UX violations Palette Not UI

Termination Conditions (Priority Order)

  1. All quality targets achieved (UQS >= target)
  2. Diminishing returns (delta < 5% for 2 consecutive cycles)
  3. Maximum cycles reached (default: 3)
  4. User manual stop

Cycle Modes

Mode Max Cycles Target UQS Use Case
QUICK 2 70 Fast turnaround
STANDARD 3 80 Balanced (default)
INTENSIVE 5 90 High-quality requirements

Source: SKILL.md on GitHub

3 warnings13d5 checks · Risk MEDIUM
  • Gen Agent Trust Hub13d

    The 'nexus' skill is a comprehensive multi-agent orchestration framework that manages complex task chains. While it incorporates extensive internal guardrails and verification protocols, it explicitly mandates the use of high-risk flags that bypass security permissions to achieve autonomy. It also provides instructions for establishing persistent tasks via cron and GitHub Actions, and utilizes external research tools to fetch content from the web.

  • Socket13d

    1 alert: gptSecurity

  • Snyk13d

    Risk: LOW · No issues

  • Runlayer6mo

    6/22 files flagged

  • ZeroLeaks5mo

    Score: 93/100 · 2 sections analyzed

Signed by skilld at c805268. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub 2 days ago.

Activeupdated 2 weeks ago

README badge

README badge for simota/agent-skills/nexus