agentleFS
MarkDonishMarkDonish/round-table-workspace/AGENTS.md

round-table-workspace

AGENTS.md2 starsChanged 7 months agoMITView on GitHub

Project entrypoint for this repository. Keep this file short and repo-local so it works the same way on Windows, Mac, and Codex Cloud. Treat this repository as the long-term source of truth for the round-table system. Primary source directories: Historical material lives under: For every new development task, follow: Default rule: Do not treat exploratory local edits as complete work. Do not treat reports/ as the active implementation source. Local Superpowers is a workflow helper, not a source-of-truth override. /room…

  • Where files are headingNew AGENTS.md files have more rules about keeping secrets out of the code, and fewer about code style.

    Of the AGENTS.md files we read, the 31,439 first seen in the last six months, against the older ones: the kind of rule that gained most share, and the kind that lost most. About files like this in general, not this one, and an estimate, since a kind is a pattern in the words.

  • In this repository3 other files it gives its agents.

The file

We split this AGENTS.md into the rules it gives an agent

9 rules. About 0 are common, 4 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈4

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Do not fabricate missing facts.

    line 7
  2. Distinctive:

    Keep outputs practical, clear, and structured.

    line 8
  3. Very specific:

    Do not enter /room or /debate unless the user explicitly starts with that command or is already continuing inside that mode.

    line 10
  4. Distinctive:

    Prefer repo-relative paths and repo-local references over machine-specific absolute home-directory paths.

    line 11
  5. Distinctive:

    Treat this repository as the long-term source of truth for the round-table system.

    line 15
  6. Very specific:

    Do not treat exploratory local edits as complete work.

    line 62
  7. Very specific:

    Do not treat reports/ as the active implementation source.

    line 63
  8. Very specific:

    Only after the user explicitly starts with /room or /debate do you enter that workflow.

    line 82
  9. Distinctive:

    Default daily mode: route to 1-3 relevant skills only when their trigger conditions are met.

    line 144

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md

Project entrypoint for this repository. Keep this file short and repo-local so it works the same way on Windows, Mac, and Codex Cloud.

Always-On Rules

  1. Do not fabricate missing facts. If information is insufficient, say 信息缺失.
  2. Keep outputs practical, clear, and structured. Avoid slogans, fluff, and unsupported certainty.
  3. Distinguish facts, inferences, uncertainties, and recommendations when doing multi-agent or multi-perspective work.
  4. Do not enter /room or /debate unless the user explicitly starts with that command or is already continuing inside that mode.
  5. Prefer repo-relative paths and repo-local references over machine-specific absolute home-directory paths.

Source Of Truth

Treat this repository as the long-term source of truth for the round-table system.

Primary source directories:

  • README.md
  • LAUNCH.md
  • AGENTS.md
  • docs/
  • schemas/
  • agents/
  • config/
  • roundtable_core/
  • scripts/
  • skills_src/
  • evals/
  • prompts/
  • examples/
  • .codex/skills/debate-roundtable-skill/
  • .codex/skills/room-skill/
  • .codex/skills/agent-builder-skill/
  • .codex/skills/*/roundtable-profile.md
  • .claude/skills/ (Claude Code project-skill adapter layer only; not protocol implementation)

Historical material lives under:

  • reports/
  • artifacts/

Development Sync

For every new development task, follow:

  • docs/source-truth-map.md
  • docs/development-sync-protocol.md
  • docs/release-readiness.md when deciding launch readiness or remaining blockers
  • docs/release-candidate-scope.md when creating a launch/support claim
  • docs/agent-continuity-checkpoints.md when host-level memory is read-only or another agent will continue the work
  • docs/superpowers/local-development-integration.md when local Superpowers is installed

Default rule:

  • develop locally
  • verify locally
  • commit verified changes
  • push to origin/main
  • report completed work, verification, and remaining gaps

Do not treat exploratory local edits as complete work. Do not treat reports/ as the active implementation source.

Local Superpowers is a workflow helper, not a source-of-truth override.

Room Roundtable

/room is an explicit-only stateful roundtable mode. It is not the default mode.

Command Discovery Hint

For complex strategy, product, risk, or decision questions, you may briefly suggest that the user can choose /room or /debate. This is only a hint, not a mode switch.

Examples:

  • Ordinary factual question: answer directly; no /room or /debate hint.
  • Complex product/startup judgment: "这个问题适合用 /room 先探索,或用 /debate 做正式判断。你可以直接输入 /room ... 或 /debate ...。"
  • Only after the user explicitly starts with /room or /debate do you enter that workflow.

On /room, use:

  • Skill: .codex/skills/room-skill/SKILL.md
  • Claude Code project skill: .claude/skills/room/SKILL.md
  • Supporting docs:
    • docs/room-runtime-status.md
    • docs/room-runtime-bridge.md
    • docs/host-adapter-architecture.md
    • docs/generic-local-agent-adapter.md
    • docs/local-agent-host-recipes.md
    • docs/third-party-agent-wrapper-recipes.md
    • docs/provider-live-readiness.md
    • docs/release-readiness.md
    • docs/release-candidate-scope.md
    • docs/claude-code-skill-adapter.md
    • docs/room-architecture.md
    • docs/room-selection-policy.md
    • docs/room-chat-contract.md
    • docs/agent-registry.md
    • docs/room-to-debate-handoff.md
    • schemas/room-session.schema.json
    • prompts/room-selection.md
    • prompts/room-chat.md
    • prompts/room-summary.md
    • prompts/room-upgrade.md

Debate Roundtable

/debate is an explicit-only roundtable dispatcher. It is not the default mode.

On /debate, use:

  • Skill: .codex/skills/debate-roundtable-skill/SKILL.md
  • Claude Code project skill: .claude/skills/debate/SKILL.md
  • Supporting docs:
    • docs/debate-runtime-bridge.md
    • docs/host-adapter-architecture.md
    • docs/generic-local-agent-adapter.md
    • docs/local-agent-host-recipes.md
    • docs/third-party-agent-wrapper-recipes.md
    • docs/provider-live-readiness.md
    • docs/release-readiness.md
    • docs/release-candidate-scope.md
    • docs/claude-code-skill-adapter.md
    • docs/debate-skill-architecture.md
    • docs/agent-role-map.md
    • docs/reviewer-protocol.md
    • docs/red-flags.md
    • docs/room-to-debate-handoff.md
    • schemas/debate-session.schema.json
    • schemas/debate-result.schema.json
    • prompts/debate-roundtable.md
    • prompts/debate-reviewer.md
    • prompts/debate-followup.md

Historical archive only when needed for archaeology, not as active source of truth:

  • docs/archive/AGENTS.full-20260411-142320.md

Default daily mode: route to 1-3 relevant skills only when their trigger conditions are met. Do not auto-run a room or roundtable.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.