Claude Skill MartBrowse skillsWhat is a Skill?
 
← Back to list

Hyperplan (Adversarial Multi-Agent Planning)

Spins up 5 mutually hostile agents for a 3-round critique debate, keeps only the defensible insights, then hands them to the `plan` agent for an executable plan.

Dev & CodingAdvanced★ 67,531⑂ 5,507AI score 7/10Last updated: Aug 9, 2026

What it does

Most planning refines the first idea. Hyperplan attacks it. The orchestrator becomes the Lead of a five-member team whose members are deliberately hostile to one another:

  • skeptic — attacks over-engineering, premature abstraction, scope creep
  • validator — hunts missed edge cases, cross-module blast radius, regression vectors
  • researcher — demands file:line or doc-URL evidence for every claim
  • architect — exposes hidden coupling, leaky abstractions, technical debt
  • creative — forces at least three alternatives before any solution is accepted

Execution runs in 7 phases: independent analysis → cross-attack → defend/refine/concede → the Lead distills survivors into Hard Constraints, Decisions, Risks & Mitigations, and Open Questions → a mandatory handoff to the plan agent for sequencing, dependencies and per-task success criteria → team cleanup.

Who it's for

  • Engineers about to start a large refactor or architecture change who want weak assumptions surfaced first
  • Teams burned by happy-path-only plans
  • Anyone uncomfortable accepting an AI's first answer as the plan
  • Advanced users already running opencode with team-mode enabled

Examples

  1. Payment module refactor: "hyperplan redesigning the payment service interface" — skeptic demands the new abstraction layer be deleted, validator flags existing webhook tests that will break, and the final plan includes a regression gate.
  2. New feature spec review: "hpp plan for realtime notifications" — creative forces a polling vs. SSE vs. WebSocket comparison, architect records the coupling-based rationale for the winner.
  3. Migration planning: "adversarial plan for the DB schema migration" — researcher strips out unevidenced performance claims, and unresolved points become explicit user-input gates in the plan.

· · · Install guide · · ·

Install in the Claude app (no terminal)
  1. Download the ZIP with the button below.
  2. In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
  3. Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
↓ Download ZIP
Install in Claude Code

Let Claude do it — paste this into Claude Code

Install the skill I found on Claude Skill Mart.
Copy the .opencode/skills/hyperplan folder from the GitHub repo code-yeongyu/oh-my-openagent into my ~/.claude/skills/code-yeongyu-hyperplan/.
When it's done, tell me in one line what this skill can do.

Install with a command instead

git clone https://github.com/code-yeongyu/oh-my-openagent.git /tmp/oh-my-openagent && mkdir -p ~/.claude/skills && cp -r /tmp/oh-my-openagent/.opencode/skills/hyperplan ~/.claude/skills/

⚠ This is a third-party skill. Check the source repository before installing.

  1. Open a terminal.
  2. Clone the repository: git clone https://github.com/code-yeongyu/oh-my-openagent.git
  3. Copy the skill folder: mkdir -p ~/.claude/skills && cp -r oh-my-openagent/.opencode/skills/hyperplan ~/.claude/skills/
  4. If you run opencode, set team_mode.enabled: true in ~/.config/opencode/oh-my-opencode.jsonc and restart opencode. Without team-mode the skill stops at its precondition check.
  5. Restart Claude Code / opencode, then invoke it with a trigger word: hyperplan a plan to refactor the payment module (also hpp, adversarial plan).
  6. You should see "HYPERPLAN MODE ENABLED!". The three debate rounds consume significant time and tokens, so let it finish until the plan agent returns the formalized plan.
View source on GitHub ↗License: NOASSERTION