refactor: rewrite bmad-party-mode as platform-adaptive multi-agent roundtable#2128
Closed
refactor: rewrite bmad-party-mode as platform-adaptive multi-agent roundtable#2128
Conversation
…undtable skill Replaces the old workflow-based party mode with an outcome-based skill that spawns independent sub-agents per BMAD agent. Adds platform adapters for Claude Code, Codex, and Gemini CLI, and restructures steps around initialize, orchestrate, and exit phases. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
🤖 Augment PR SummarySummary: This PR rewrites Changes:
Technical Notes: The orchestration stage now maintains a compact state block to survive context compaction and adapts behavior based on momentum/quality signals across rounds. 🤖 Was this summary useful? React with 👍 or 👎 |
![augmentcode[bot]](https://avatars.githubusercontent.com/in/1027498?s=60&v=4){kind=small}
📝 WalkthroughWalkthroughThis PR comprehensively restructures the BMAD Party Mode skill documentation, replacing a minimal workflow specification with detailed platform-specific orchestration guidelines for Claude Code, Codex, and Gemini CLI, introducing three-stage step workflows (initialize, orchestrate, exit), agent prompt templates, and removing legacy workflow files. Changes
Estimated code review effort🎯 4 (Complex) | ⏱️ ~40 minutes Possibly related PRs
Suggested reviewers
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches📝 Generate docstrings
🧪 Generate unit tests (beta)
Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 3
🧹 Nitpick comments (1)
src/core-skills/bmad-party-mode/steps/step-03-exit.md (1)
20-30: Farewell prompt contract is duplicated and already drifted from the shared template.This block diverges from
references/agent-prompt-template.md(different placeholder schema), so maintenance will fork behavior. Prefer referencing the shared “Farewell Variant” contract directly.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/core-skills/bmad-party-mode/steps/step-03-exit.md` around lines 20 - 30, The farewell block using placeholders like {displayName}, {title}, and {personality_profile} duplicates and diverges from the shared "Farewell Variant" contract; remove this duplicated block in step-03-exit.md and replace it by referencing or importing the canonical Farewell Variant template (use the shared template's placeholder schema), ensuring the output starts with the required token (e.g., the shared template's leading "{icon} **{displayName}**:" or its canonical equivalent) and that no custom placeholder names remain so behavior follows the single source of truth.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/core-skills/bmad-party-mode/adapters/codex.md`:
- Around line 35-43: The Codex agent template's developer_instructions block is
missing the explicit no-tools constraint, so update the codex.md
developer_instructions to include a clear "do not call or invoke any external
tools" (or "no_tools") rule so sub-agents cannot use read-only tools and will
adhere to party-mode; locate the developer_instructions section in
src/core-skills/bmad-party-mode/adapters/codex.md and add a concise sentence
such as "Do not call, invoke, or rely on external tools or APIs (no tools)"
within the instruction list for the Codex agent template to enforce the
constraint.
In `@src/core-skills/bmad-party-mode/references/agent-prompt-template.md`:
- Around line 20-22: The template placeholder in
src/core-skills/bmad-party-mode/references/agent-prompt-template.md is
inconsistent with the orchestrator: replace the token
`{pass_1_responses_if_cross_talk_pass — empty on first pass}` with
`{agent_responses_from_pass_1}` so it matches the token used in
src/core-skills/bmad-party-mode/steps/step-02-orchestrate.md (lines referencing
agent_responses_from_pass_1); ensure any other occurrences of the old token are
updated to the same `{agent_responses_from_pass_1}` name so template assembly
works end-to-end.
In `@src/core-skills/bmad-party-mode/steps/step-02-orchestrate.md`:
- Line 121: The "Redundancy" pass-1 rule in step-02-orchestrate.md conflicts
with the party-mode hard constraint by allowing collapsing of pass-1 responses;
change the rule so it does not collapse or summarize any pass-1 outputs—ensure
all initial responses are emitted in full (no summarization/merging) before any
cross-talk or aggregation logic runs, and update the wording for the
"Redundancy" bullet to explicitly preserve full individual pass-1 responses per
the party-mode hard constraint.
---
Nitpick comments:
In `@src/core-skills/bmad-party-mode/steps/step-03-exit.md`:
- Around line 20-30: The farewell block using placeholders like {displayName},
{title}, and {personality_profile} duplicates and diverges from the shared
"Farewell Variant" contract; remove this duplicated block in step-03-exit.md and
replace it by referencing or importing the canonical Farewell Variant template
(use the shared template's placeholder schema), ensuring the output starts with
the required token (e.g., the shared template's leading "{icon}
**{displayName}**:" or its canonical equivalent) and that no custom placeholder
names remain so behavior follows the single source of truth.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
Run ID: e5f8b97d-9fde-4e58-8102-df54f6f39412
📒 Files selected for processing (12)
src/core-skills/bmad-party-mode/SKILL.mdsrc/core-skills/bmad-party-mode/adapters/claude-code.mdsrc/core-skills/bmad-party-mode/adapters/codex.mdsrc/core-skills/bmad-party-mode/adapters/gemini.mdsrc/core-skills/bmad-party-mode/references/agent-prompt-template.mdsrc/core-skills/bmad-party-mode/steps/step-01-agent-loading.mdsrc/core-skills/bmad-party-mode/steps/step-01-initialize.mdsrc/core-skills/bmad-party-mode/steps/step-02-discussion-orchestration.mdsrc/core-skills/bmad-party-mode/steps/step-02-orchestrate.mdsrc/core-skills/bmad-party-mode/steps/step-03-exit.mdsrc/core-skills/bmad-party-mode/steps/step-03-graceful-exit.mdsrc/core-skills/bmad-party-mode/workflow.md
💤 Files with no reviewable changes (4)
- src/core-skills/bmad-party-mode/workflow.md
- src/core-skills/bmad-party-mode/steps/step-02-discussion-orchestration.md
- src/core-skills/bmad-party-mode/steps/step-03-graceful-exit.md
- src/core-skills/bmad-party-mode/steps/step-01-agent-loading.md
src/core-skills/bmad-party-mode/references/agent-prompt-template.md
Outdated
Show resolved
Hide resolved
Inline all adapters, steps, references, and prompt templates directly into SKILL.md. Remove the platform-adapter detection system in favor of a simpler approach using the Agent tool. Streamline agent selection from a weighted scoring algorithm to pragmatic guidelines. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
bmadcode
requested changes
Mar 29, 2026
| - Do NOT use tools. Just respond with your perspective. | ||
| ``` | ||
|
|
||
| **Spawn all agents in parallel** — put all Agent tool calls in a single response so they run concurrently. Use `model: "sonnet"` for the subagents to keep things fast. |
Collaborator
There was a problem hiding this comment.
we cannot do this it needs to work across tools and models. we could offer a subagent model commandline arg though. but then only use it if available.
Another option is in the future, we could let the user let us know what models they have registered or available in some config and we could allow the user to tell us what to use as a fast agent, worker agent and thinking agent for example. but beyond the scope of this.
Contributor
Author
There was a problem hiding this comment.
alright made changes per our discussion and removed bmad-init dependency
Allow users to override the subagent model and run in solo mode (single-LLM roleplay) when subagents aren't available or speed is preferred. Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
…party-mode Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
bmad-party-modeskill from a simple workflow delegation into an outcome-based skill that spawns independent sub-agents per BMAD agentTest plan
🤖 Generated with Claude Code