Runs iterative consensus planning with Planner, Architect, and Critic agents until plan approval.
npx skills add https://github.com/l0lxl0lw/dotfiles --skill ralplan-l0lxl0lw
Please help me install this Agent Skill. Skill: ralplan Source: https://github.com/l0lxl0lw/dotfiles/tree/main/ai/shared/skills/omc/ralplan Command: npx skills add https://github.com/l0lxl0lw/dotfiles --skill ralplan-l0lxl0lw
/plan --consensus to cycle Planner, Architect, and Critic agents through up to 5 review iterations until the Critic approves. - RALPLAN-DR Structured Deliberation: Produces principles, decision drivers, viable options with tradeoffs, and an ADR; deliberate mode adds pre-mortem scenarios and expanded test planning for high-risk work. - Interactive or Automated Modes: Use --interactive for user checkpoints at draft review and final approval, or run fully automated end to end. - Use Case: Before migrating a production database schema, run /ralplan --deliberate "migrate users table to new schema" to get a plan vetted by architectural review, risk analysis, and quality critique. ## Quick Start Ask the AI to run /ralplan with a description of the task you want planned, optionally adding --interactive or --deliberate flags.High-intent search queries and answers about installing and using this skill.
Run /ralplan followed by your task description to invoke /plan --consensus. The Planner drafts a plan, the Architect reviews it with a steelman antithesis, and the Critic evaluates quality, looping up to 5 times until approval.
Ralplan is an alias for /plan --consensus, meaning it always runs the multi-agent consensus workflow with Planner, Architect, and Critic. The base plan skill supports other modes; ralplan fixes the consensus mode and adds RALPLAN-DR structured deliberation.
Use --deliberate for high-risk work such as auth or security changes, migrations, destructive operations, production incidents, compliance or PII handling, and public API breakage. It adds a 3-scenario pre-mortem and expanded unit, integration, e2e, and observability test planning.
Yes, without the --interactive flag the workflow runs fully automated from Planner through Architect and Critic loops and outputs the final plan. With --interactive, it prompts you at draft review and final approval using AskUserQuestion.
The re-review loop runs a maximum of 5 iterations of Architect and Critic feedback with Planner revisions. If no APPROVE verdict is reached after 5 iterations, the best version of the plan is presented to the user.
henrycashe26Generate structured Product Requirement Documents from conversation context and codebase analysis.
swissarmyhammerAutomates sequential processing of all IT service management tasks from kickoff to completion.
PaloAltoNetworksAutomate architectural guardrail, drift, and ADR impact validation on code changes.
21StarkComAuthors human-gated specs and task DAGs through recon, interview, and sign-off.