agora

Agora

English | 한국어 | 中文 | 日本語

Agents generate options. Agora helps humans supervise judgment.

Agora is a skill-first workflow registry for supervised AI judgment in Claude Code and Codex. It adds clarification, doubt, dissent, synthesis, and governance workflows to agent work that needs reviewable artifacts instead of one-off answers.

Agora provides reusable reasoning commands, durable judgment artifacts, workflow registry routing, and philosophical modes when a problem needs more than raw generation. Unsupported or pending providers are intentionally not listed in the quick start.

Installation quick start

Use the verified surfaces below.

Claude Code

# Verified Git marketplace path
/plugin marketplace add malleus35/agora
/plugin install agora@malleus35-agora

skills.sh CLI

npx skills add https://github.com/malleus35/agora --skill agora

Codex

Fetch and follow instructions from https://raw.githubusercontent.com/malleus35/agora/refs/heads/main/.codex/INSTALL.md

Quick start

The easiest way to use Agora is to start from the common entrypoint and describe your problem naturally.

Claude Code

/agora "your problem"

Codex

$agora "your problem"

If you want to use a specific workflow directly, you can still call the underlying skill or command yourself. For example: clarify-goals, cartesian-grill, doubt-list, or court-review.

What changed in v2

Agora is no longer positioned as a debate harness first. It is now an overlay that helps teams:

• clarify the real decision before acting
• preserve dissent instead of flattening it
• write doubt lists before trusting outputs
• grill plans one question at a time before batch verification
• surface hidden assumptions and inherited categories
• produce reviewable artifacts for human approval

Philosophy is methodology, not decoration. Dialectic, skepticism, genealogy, and court review are practical modes, not roleplay.

Core skills, modes, and workflows

Core skills

Small reusable workflows that can be used independently:

• agora
• clarify-goals
• cartesian-grill
• prd-from-requirements
• tdd-subagent-implementation
• frame-the-decision
• compare-options
• doubt-list
• assumption-audit
• minority-report
• decision-memo
• steelman
• synthesis-memo
• court-review

Modes

Opinionated reasoning modes for problems that need a stronger worldview. Modes are not installed as skills:

• dialectic — strategic synthesis and ideation
• skeptic — verification and risk review
• genealogy — assumption and value critique
• court — governance and approval structure

Workflow registry

agora contains a workflow registry that keeps multi-skill flows connected even when host skill chaining is unreliable. See docs/workflows.md.

Commands

Agora exposes small entrypoints so users can get value from one command without learning the whole system.

• /agora "prompt" — route to the right workflow
• /clarify "prompt" — extract purpose, constraints, success criteria, unknowns
• /decide "prompt" — structure a real fork and write a decision memo
• /doubt "prompt" — generate a Descartes-style doubt list and verification plan
• /dialectic "prompt" — run explicit Hegelian synthesis mode for contradiction and ideation
• /steelman "prompt" — restate an opposing view in strongest form before critique
• /assumption-audit "prompt" — expose hidden premises, inherited categories, and value conflicts
• /minority-report "prompt" — preserve unresolved objections and revisit triggers
• /court-review "prompt" — separate strategist, operator, censor, historian, sovereign views

Quick examples

/agora "We have an architecture disagreement and need the right reasoning mode."

/clarify "We need an internal agent harness, but I am not sure whether the first goal is reliability or collaboration."

/decide "Should our agent harness use a shared-spec overlay or a repo-per-agent strategy?"

/doubt "Review this release plan for hidden failure modes before we ship."

/dialectic "We need synthesis between low-friction adoption and philosophical differentiation."

/agora "Grill this PRD before implementation and tell me what still needs verification."

How /agora routes work

• vague problem -> clarify-goals
• real decision fork -> frame-the-decision + compare-options
• plan / design / PRD stress test -> plan-stress-test workflow: clarify-goals + cartesian-grill + doubt-list
• clear requirements -> prd-from-requirements
• implementation -> tdd-subagent-implementation
• open-ended ideation -> dialectic mode + synthesis workflow
• verification / risk / pre-release review -> skeptic mode + doubt-list
• governance / approval / responsibility separation -> court mode + court-review

Dialectic remains important, but it is no longer the default for every task.

Artifacts Agora produces