By carbonshow
Orchestrate autonomous AI agents that transform vague intents into fully delivered projects through iterative research, adversarial review, slide presentations, and multi-round development cycles from PRDs or specs.
Use proactively when the user needs to make sense of an unfamiliar, ambiguous, complex, or messy domain before acting. Trigger not only on explicit requests about 快速掌握未知领域, 深度调研, 调研方法论, exploratory search, sensemaking, knowledge graph learning, frontier exploration, research workflow, evidence synthesis, or industry consulting, but also on vague questions, 'I do not know where to start', learning plans, market/industry/technology/product trend analysis, competitor/user/interview/observation synthesis, scientific or engineering investigation design, hypothesis generation, theory building, decision framing, strategy memo creation, and situations where the task requires iteratively reframing the question, ranking what to explore next, building concept/claim/evidence relationships, or deriving an actionable conclusion from incomplete information. Do not use for simple fact lookup, narrow one-shot Q&A, or tasks with an already fixed implementation path.
Use when a user asks to evaluate a complex, high-impact, ambiguous, cross-functional, or hard-to-reverse decision; mentions expert panel, red team, adversarial review, rebuttal, judge, arbitration, risk review, challenge assumptions, poke holes, or wants stronger decision quality than a normal review.
Use when the user wants to create slides, build a presentation or talk, make a deck or pitch, export slides to PDF, work with an existing slides.md, run slidev dev mode, or build a static slide site. Also applies when the user mentions "presentation", "keynote", "talk", "lecture", "deck", "slide deck", or asks to turn content into slides.
Use when a user provides a PRD, spec, or detailed requirements document and needs a full project delivered through iterative expert orchestration — multi-round analyze/research/design/implement/QA cycles with convergence detection. NOT for: single-file edits, quick prototypes, simple Q&A, or tasks without a written spec.
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Transforming human intent into autonomous execution.
Stop manual grinding, start directing intelligence.
Your browser does not support the video tag. Download the demo video.
One PRD in → complete deliverables out. Watch Surge autonomously iterate through Analyze → Research → Design → Implement → QA until quality converges.
🔄 Autonomous Iteration Engine Director Agent drives Analyze → Research → Design → Implement → QA in a closed loop. Not a one-shot generator — it iterates until quality converges. Auto-detects stagnation, oscillation, and Pareto frontiers to reach highest quality in minimum iterations.
👥 Expert Review Panel Automatically assembles 3-5 domain experts for parallel design review. Each expert evaluates independently — security, performance, maintainability — with veto power. Multi-perspective synthesis eliminates blind spots. High-risk decisions can optionally use expert-redteam-review for adversarial arbitration.
✅ Rigorous Quality Assurance Three-tier acceptance criteria (L1→L2→L3) with progressive escalation. Output integrity validation auto-detects truncation and recovers. Optimization directives are closed-loop tracked — every improvement proposed is verified next round.
📋 PRD-to-Deliverable Pipeline One PRD in, complete deliverables out — code, documents, or strategy reports. Auto-analyzes requirement topology, negotiates acceptance criteria, orchestrates parallel subtasks. From fuzzy intent to structured output, fully autonomous.
🧠 Self-Evolving Process Memory Extracts process experience after every iteration — ambiguities found, reusable components, rejected approaches, missing test cases. Persisted to memory files — gets smarter with every use. Retro phase generates rule update suggestions.
Intent-Fluid transforms intent into structured reality through an iterative, agentic pipeline. It doesn't generate once and stop — it loops through Analyze → Research → Design → Implement → QA until quality converges on your acceptance criteria. Each iteration extracts process memory, making the next run smarter.
See the full Installation Guide for platform-specific instructions.
git clone https://github.com/carbonshow/intent-fluid.git
| Platform | Integration |
|---|---|
| Claude Code | /plugin marketplace add carbonshow/intent-fluid |
| Cursor | Add https://github.com/carbonshow/intent-fluid as a rule source or MCP server |
| Gemini CLI | gemini extensions install https://github.com/carbonshow/intent-fluid |
npx claudepluginhub carbonshow/intent-fluid --plugin intent-fluidCore skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Harness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Plugin-safe Claude Code distribution of Antigravity Awesome Skills with 1,561 supported skills.
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
v9.44.1 — Patch release for Gemini environment/version detection and qwen auth gating. Run /octo:setup.