hashgraph-online/awesome-codex-plugins / plugins/beastlabai/multi-llm-plugin/skills/multi-llm
multi-llm
FlaggedOrchestrate plan reviews, task generation, implementation and code reviews across several LLM CLIs in parallel, or ask every model one free-text question about a plan. Use when the user says: review my plan, multi-llm, run a code review, generate tasks from a plan, implement this plan, ask the models about my plan.
prompts/consolidate_suggestions.txt
View on GitHubYou are consolidating suggestion groups from a multi-LLM plan review.
## Input
You are given a set of suggestion groups for a specific plan section. Each group was identified by one or more LLM reviewers as a potential improvement to the plan. Many groups may address the same underlying concern with different wording or emphasis. Your job is to cluster related groups and produce a single consolidated suggestion per cluster.
### Input JSON
{GROUPS_JSON}
Total input groups: {GROUP_COUNT}
## What to Do
1. Read each input group carefully, focusing on the `theme`, `importance`, and individual `suggestions[].description` fields.
2. Identify groups that address the **same underlying concern** — even if they use different wording, focus on different facets of the same issue, or propose different solutions to the same problem.
3. For each cluster of related groups, produce a **single consolidated suggestion** with:
- A new synthesized title (not just concatenating originals). Max 120 characters.
- A synthesized description that preserves nuance from ALL underlying suggestions. Mention specific points from each underlying group — do not summarize generically.
- The highest importance level among the underlying groups (HIGH > MEDIUM > LOW).
- The most actionable type among the underlying groups. Priority: `addition` > `modification` > `deletion` > `clarification`.
- The exact `group_id` values and `index` values from the input groups.
- A 1-2 sentence reasoning explaining why these groups belong together.
4. Groups that do not relate to any other group are **singletons**. Include them in the `singletons` array with their `group_id`, `group_index`, and a brief reasoning for why they stand alone.
## Output Format
Output ONLY valid JSON matching this exact structure:
{OUTPUT_SCHEMA}
## Example
### Example Input (3 groups):
{EXAMPLE_INPUT}
### Example Output:
{EXAMPLE_OUTPUT}
## Critical Rules
{NON_OVERLAP_CONSTRAINTS}
## Ambiguous Case Handling
- **Group fits multiple clusters**: Assign to the cluster where the overlap is strongest (most shared concepts). In your `reasoning`, note the secondary cluster connection: "Also related to [other cluster topic] but primary concern is [this cluster topic]."
- **All groups are unrelated**: Return an empty `clusters` array and all groups as `singletons`. This is a valid outcome.
- **All groups are related**: Return a single cluster containing all groups. This is valid but should only happen if they genuinely address the same underlying concern.
- **Borderline cases**: When uncertain whether two groups address the same concern, prefer keeping them separate (as singletons or smaller clusters). False merges are worse than missed merges — users can always mentally group singletons, but splitting a bad merge requires re-running consolidation.
## Assignment Rule
ASSIGNMENT RULE: Each input group must appear in EXACTLY ONE cluster's underlying_group_ids OR in singletons. Never in multiple clusters. Never in both a cluster and singletons. Never omitted.
## Output
Output ONLY the JSON object described above. No markdown, no code blocks, no explanation — just raw JSON.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| receiving-code-reviewobra/superpowersUse when receiving code review feedback, before implementing suggestions, especially if feedback seems unclear or technically questionable - requires technical rigor and verification, not performative agreement or blind implementation | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| using-git-worktreesobra/superpowersUse when starting feature work that needs isolation from current workspace or before executing implementation plans - ensures an isolated workspace exists via native tools or git worktree fallback | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| verification-before-completionobra/superpowersUse when about to claim work is complete, fixed, or passing, before committing or creating PRs - requires running verification commands and confirming output before making any success claims; evidence before assertions always | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |