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.
schemas/generated_tasks.schema.json
View on GitHub{
"$schema": "http://json-schema.org/draft-07/schema#",
"$id": "generated_tasks.schema.json",
"title": "Task-generation result document (task_generation_result.json)",
"description": "The document a task_generation subagent prints and the invoker persists under the phase dir. Validated before anything is written to the plan; the semantic rules utils/task_decomposer.py owns (unique T### ids, depends_on naming only declared ids, acyclic, at least one dependency-free task) run in Python immediately after this schema check. A document that fails either is a schema_invalid failure; the executor lane re-asks the runner once with the validator's errors (the retry ladder's rule 7), and nothing else retries it.",
"type": "object",
"additionalProperties": false,
"required": ["schema_version", "plan_preamble", "tasks"],
"properties": {
"schema_version": {"type": "string", "const": "1.0"},
"plan_preamble": {
"type": "string",
"minLength": 1,
"description": "3-5 sentences of shared context injected into every task prompt. Required, and non-empty: the task_generation reply contract emits it unconditionally, and a document that omits it would validate cleanly while every downstream implement prompt silently loses its shared context."
},
"tasks": {
"type": "array",
"minItems": 1,
"items": {"$ref": "#/$defs/generated_task"}
}
},
"$defs": {
"generated_task": {
"type": "object",
"additionalProperties": false,
"required": [
"id",
"title",
"description",
"depends_on",
"files_to_create",
"files_to_modify",
"acceptance_criteria",
"estimated_complexity",
"executor_role",
"subagent_type"
],
"properties": {
"id": {"type": "string", "pattern": "^T[0-9]{3}$"},
"title": {"type": "string", "minLength": 1},
"description": {"type": "string", "minLength": 1},
"depends_on": {"type": "array", "items": {"type": "string"}},
"files_to_create": {"type": "array", "items": {"type": "string"}},
"files_to_modify": {"type": "array", "items": {"type": "string"}},
"acceptance_criteria": {"type": "array", "items": {"type": "string"}},
"estimated_complexity": {"type": "string", "enum": ["low", "medium", "high"]},
"executor_role": {"type": "string", "enum": ["agent", "human"]},
"subagent_type": {
"type": ["string", "null"],
"enum": ["general-purpose", "human", null],
"description": "Optional Claude-Code-only metadata, so required-but-nullable: OpenAI's strict structured-output subset needs every property in `required` (measured 2026-10-01: codex declined this schema on `subagent_type` alone), and null is how a reply says it has none. The invoker supplies null when a reply leaves the key out (subagent_executor.normalize_result). When non-null it must agree with executor_role; that rule lives in task_decomposer.validate_generated_tasks because OpenAI's strict subset does not list allOf/if/then (the former allOf here is gone, not weakened)."
}
}
}
}
}
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: 321k | Updated | Claude CodeCodex | Verified | Install |
| requesting-code-reviewobra/superpowersUse when completing tasks, implementing major features, or before merging to verify work meets requirements | Stars: 321k | 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: 321k | 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: 321k | Updated | Claude CodeCodex | Verified | Install |