Help us improve
Share bugs, ideas, or general feedback.
From erpaval
Generator-critic planning: launches 3 parallel explorers with distinct biases (architectural, speed-first, simple-first), then composes best elements into a single plan with attribution. For non-trivial planning tasks.
npx claudepluginhub theagenticguy/erpaval --plugin erpavalHow this skill is triggered — by the user, by Claude, or both
Slash command
/erpaval:ultraplanThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
| Reference | When to load |
Refines development plans through multi-round discussions among 3-10 specialized agents selected by domain like frontend, backend, security. Builds consensus on recommendations and trade-offs.
Creates structured plans for multi-step tasks including software features, implementations, research, or projects. Deepens plans via interactive sub-agent reviews.
Strategic planning with auto-calibrated detail, decision rationale, and dependency ordering. Use when starting a new feature, bug fix, refactor, or any non-trivial work. Produces a plan document with tasks, reasoning, and acceptance criteria. Triggers: plan, planning, create plan, implementation plan, feature plan, work plan.
Share bugs, ideas, or general feedback.
| Reference | When to load |
|---|---|
references/orchestrator.md | Running /ultraplan <problem> — the 5-phase flow |
references/divergence-vectors.md | The three explorer biases and their motivation |
references/write-protocol.md | Copy verbatim into every explorer prompt |
templates/explorer-skeleton.md | Per-explorer plan skeleton |
templates/plan-skeleton.md | Final composed plan skeleton |
Generator-critic planning, file-coordinated. Three Opus 4.7 explorers plan the same problem in parallel, each biased toward a different axis — architectural, speed-first, simple-first. A critic reads all three and composes a final plan that cherry-picks the best decision from each, with attribution. The user gets one clean plan.md to execute against.
Beats single-agent planning on two axes:
Read references/orchestrator.md and follow its phases. Short version:
AskUserQuestion only if scope is genuinely ambiguous; otherwise proceed.planning/{slug}/ with three explorer files and a plan.md shell.general-purpose Opus 4.7 Agents in parallel, run_in_background: true, each with a distinct divergence vector from references/divergence-vectors.md. Orchestrator does not read the codebase; explorers do.wc -l on explorer files; stuck detection via line-count delta; launch a fresh Agent with "resume from section N" context on stall.general-purpose Opus 4.7 Agent reads all three explorer files and writes plan.md — chosen approach, per-decision attribution, explicit tradeoffs, step-by-step order, verification criteria.Every explorer prompt carries the write protocol verbatim from references/write-protocol.md: think → edit the file with the decision → next consideration. Partial plans written to disk survive timeouts; plans held in working memory do not.
Three parallel Opus 4.7 explorers plus one critic synthesis. Significant token spend per run — worth it for meaningful planning decisions, not for trivial refactors. The user invoked /ultraplan because they want the divergence; ship it.