skills.rest
nyldn
nyldnCOMMUNITY
Explore all skills by this author →
Website𝕏@nyldnGitHub

council-verdicts

Interpret council run verdict artifacts including quorum, dissent, and cross-lab validity.

4.0k|375|Updated Jan 15, 2026
Search Tags:#consensus#council#multi-llm#decision review#quorum#verdict interpretation#dissent analysis
One-click install
npx skills add https://github.com/nyldn/claude-octopus --skill council-verdicts
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: council-verdicts
Source: https://github.com/nyldn/claude-octopus/tree/main/skills/octopus-starter-pack/council-verdicts
Command: npx skills add https://github.com/nyldn/claude-octopus --skill council-verdicts

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

After a multi-provider council run finishes, its raw artifacts (summary.json and per-seat verdict files) are hard to act on. This Skill reads those artifacts and explains what the verdicts actually mean before anyone makes a decision based on them.

Core Features & Use Cases

  • Artifact Location and Validation: Finds the newest council output under ~/.claude-octopus/results/ and confirms each seat's real provider and model, flagging same-lineage panels that weaken verdict independence.
  • Verdict Tallying and Quorum Checks: Counts APPROVE, REJECT, ABSTAIN, and REVISE votes, treats malformed verdicts as abstentions, and checks the SubagentStop gate's quality scores.
  • Dissent Analysis and Action Mapping: Quotes the strongest dissenting argument verbatim and maps the tally to a concrete next step, such as proceeding, proceeding with follow-ups, or re-running.
  • Use Case: After a /octo:council run on an architecture decision, ask what the council decided and receive a plain-language verdict summary with quorum status, dissent highlights, and a recommended action.

Quick Start

Ask the assistant to interpret the latest council run results and explain what the verdicts mean for your pending decision.

Frequently Asked Questions about council-verdicts

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

FAQPage Schema
How do I interpret the results of a multi-model council run?▼

Locate the newest council output under ~/.claude-octopus/results/ and read summary.json plus the per-seat verdict files. Tally the APPROVE, REJECT, ABSTAIN, and REVISE votes, check quorum, and review any dissenting arguments before acting.

What does a 2-1 council verdict with dissent mean?▼

A 2-1 split with a substantive dissent is a weaker mandate than a unanimous 3-0 approval. The recommended action is to proceed but track the dissent's concern as an explicit follow-up item rather than ignoring it.

Why does seat model lineage matter in council verdicts?▼

Two seats backed by the same model family weaken the verdict's independence because they may share blind spots. The skill flags same-lineage panels by checking each seat's resolved provider and model in summary.json.

What happens when a council seat returns a malformed verdict?▼

A malformed or missing verdict counts as an ABSTAIN in the tally. The SubagentStop gate (hooks/subagent-stop-gate.sh) may have flagged it, so check the gate's usage log for quality scores before concluding.

When should a council run be re-run instead of acted on?▼

Re-run when seats disagree on facts rather than judgment, since factual disputes make the run inconclusive. Resolve the factual question first, then run the council again for a valid verdict.

Related Skills

Explore More Software Engineering Skills→
mbriggsymbriggsy

playtest-run

Orchestrate BURNED playtest harness sessions across multiple Chromium instances.

Community
Advanced
PVPhaPVPha

remmerdoc

Creates post-task documentation with rationale, validation, and follow-ups in a structured markdown template.

Community
Basic
scanadyscanady

strategy-decision-council

Convene a four-voice decision council to surface dissent and tradeoffs.

Community
Advanced
heavy3-aiheavy3-ai
45

h3

Automates multi-model code review feedback using OpenRouter with web search and 200K token context.

Official
Advanced