By Lumio18
Turn a meeting transcript into a deployable n8n automation spec — a 7-skill pipeline covering gap detection, process inventory, opportunity matrix, stack profiling, universal automation spec, and importable n8n workflow.
Designs a universal, builder-agnostic automation spec from a prioritized opportunity (output of process-challenger) and a stack profile (output of stack-profiler). Produces a framing message, a plain-language end-user summary, a Mermaid flowchart, a universal JSON spec, deployer notes and confidence flags — in the operator's language. Trigger when both a stack profile (schema v1.1+) AND a prioritized opportunities JSON are present and the operator asks to design, architect, spec out, or shape an automation. Always proposes a recommended builder PLUS at least one alternative — never decides unilaterally. Selects the builder by priority order: existing platform in stack → tech comfort → budget → compliance → deployer skills. Inherits and surfaces every upstream confidence flag. Do NOT use to extract or model processes (process-extractor, process-modeler), to challenge them (process-challenger), to profile a stack (stack-profiler), or to generate platform-specific workflow files (n8n-builder, make-builder, …). The architect says WHAT, builders say HOW.
First skill in the Talk2Flow pipeline. Two modes — (1) guided interview from scratch: leads the end user or a consultant through a structured conversation about daily work and produces a clean transcript ready for process-extractor; (2) transcript review: accepts an existing transcript, scans it for coverage gaps (missing volumes, unnamed tools, vague pain points, actor gaps), and asks targeted complementary questions to enrich the data before extraction. Mode 2 is the default when a transcript is provided. Use mode 1 when starting from zero. Do NOT use to extract processes (that's process-extractor), model them (process-modeler), or challenge them (process-challenger). Adapts output language to the operator's language.
Translates a universal automation spec (output of automation-architect) into an importable n8n workflow JSON, an end-user summary, and a deployer handoff document. Trigger when an operator has an automation-architect spec with recommended_builder set to "n8n-builder" and wants to produce the deployable artifact. Validates the spec against schema v1.0, maps each step to a concrete n8n node, assembles the workflow JSON, and scales deployer documentation verbosity based on deployer_skills_estimated. Adapts all free-text output to the operator's language. Do NOT use to design the automation logic (that's automation-architect), or to generate Make.com / Zapier / Power Automate equivalents (those are make-builder, zapier-builder, power-automate-builder — planned V2).
Critically audits an enriched JSON produced by process-modeler to identify, qualify, and prioritize process optimization opportunities. Produces an effort/impact matrix, a four-quadrant classification (quick wins, strategic projects, marginal optimizations, time traps), and an estimated cumulative annual time saving. Trigger when an operator provides a JSON with a modeling section and wants to challenge, audit, optimize, identify time waste, or find improvement opportunities. NEVER proposes a single solution — always multiple paths (removal, standardization, existing tool, classic automation, AI, redesign). The choice of solution is the role of automation-architect that follows. Do NOT use to model processes (process-modeler) or to design automations (automation-architect). Adapts its output language to the input JSON's language.
Exhaustively and faithfully extracts the business processes described in an interview transcript or meeting note, and produces a structured JSON inventory consumable by other Talk2Flow BPM skills. Trigger when an operator provides a transcript, a meeting recap, a "day-in-the-life" interview note, or any oral description of business activity and wants to extract the underlying processes, daily tasks, or activities. Trigger even when the operator doesn't say "extract processes" explicitly — any request to analyze, structure, or summarize a business transcript should route through this skill. Do NOT use to formalize processes in BPMN (that's process-modeler), to judge value-added (that's process-challenger), or to design automations (that's automation-architect). Adapts its output language to the language used in the transcript.
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.
Turn talk into deployable workflows.
From a meeting transcript to an automation spec your developer can deploy — without consultants, without code.
Most automation projects fail before they start. A business owner knows their work is repetitive. A consultant interviews them. Notes get lost. The eventual solution doesn't fit the client's actual tools, budget, or skills. Nobody deploys it.
Talk2Flow fixes the front half of that loop.
You — or your client — describe a daily routine out loud. Talk2Flow reads the transcript, asks targeted follow-up questions to fill gaps, extracts the underlying processes, maps the existing stack, identifies optimization opportunities, and produces an automation spec adapted to what the business actually has and can run. A developer (or capable freelancer) takes that spec to production.
The last technical mile still needs a human. Talk2Flow makes that mile short.
flowchart TD
Input([Transcript or live interview])
Output([Deployer ships the workflow])
Input --> A[<b>interview-guide</b><br/>Gap detection + complementary Q&A]
A -->|enriched transcript| B[<b>process-extractor</b><br/>Process inventory JSON]
B -->|inventory| C[<b>process-modeler</b><br/>BPMN · Mermaid · SIPOC · RACI]
B -->|inventory| D[<b>process-challenger</b><br/>Opportunity matrix · ROI]
C -->|diagrams + RACI| E[<b>stack-profiler</b><br/>Tools · budget · compliance]
D -->|prioritized opportunities| E
E -->|stack profile| F[<b>automation-architect</b><br/>Universal spec — WHAT]
F -->|spec| G[<b>n8n-builder</b><br/>Importable workflow — HOW]
G --> Output
classDef io fill:#e8f4ff,stroke:#0366d6,stroke-width:2px,color:#000
classDef skill fill:#ffffff,stroke:#586069,color:#000
class Input,Output io
class A,B,C,D,E,F,G skill
Every skill is a Claude skill — a structured prompt with a clear contract, runnable in Claude.ai today. The pipeline can be run end-to-end or one stage at a time.
New in V1: when you paste a transcript, Talk2Flow automatically scans it for coverage gaps (missing volumes, unnamed tools, vague pain points) and asks targeted complementary questions before extracting — making the output dramatically more accurate.
| Generic automation builders | Workflow templates | Talk2Flow | |
|---|---|---|---|
| Starts from your actual words | ❌ | ❌ | ✅ |
| Asks follow-up questions to fill gaps | ❌ | ❌ | ✅ |
| Maps your real existing stack | ❌ | partial | ✅ |
| Adapts to your budget + skills | ❌ | ❌ | ✅ |
| Produces a deployable artifact | ✅ | ✅ | ✅ |
| Explicit handoff doc for your developer | ❌ | ❌ | ✅ |
Talk2Flow doesn't compete with n8n, Make, or Zapier. It produces the spec that feeds into them.
Three roles around any Talk2Flow project:
The framework adapts based on who is operating it. See docs/roles-and-vocabulary.md.
| Skill | What it does |
|---|---|
interview-guide | Guided interview (Mode 1) or transcript review with gap detection + follow-up Q&A (Mode 2) |
process-extractor | Transcript → structured process inventory (JSON). Multilingual. |
process-modeler | Inventory → BPMN, inline Mermaid, SIPOC, RACI, interactive BPMN viewer |
process-challenger | Inventory → effort/impact opportunity matrix, ROI estimates, dependency map |
stack-profiler | Guided interview → stack profile (tools, budget, compliance, deployment context) |
automation-architect | Opportunity + stack profile → builder-agnostic technical spec |
| Skill | Status |
|---|---|
n8n-builder | Implemented — full test on a real scenario pending |
flowchart LR
A[<b>automation-architect</b><br/>universal spec]
A ==> B[<b>n8n-builder</b><br/>V1 ✓]
A -.-> C[make-builder]
A -.-> D[zapier-builder]
A -.-> E[power-automate-builder]
A -.-> F[python-builder]
A -.-> G[your-builder<br/><i>contribute</i>]
classDef done fill:#dcffe4,stroke:#28a745,stroke-width:2px,color:#000
classDef root fill:#e8f4ff,stroke:#0366d6,stroke-width:2px,color:#000
classDef planned fill:#ffffff,stroke:#586069,stroke-dasharray:4 4,color:#000
class A root
class B done
class C,D,E,F,G planned
npx claudepluginhub lumio18/talk2flow --plugin talk2flowA growing collection of Claude-compatible academic workflow bundles. Covers scientific figures, manuscript writing and polishing, reviewer assessment, citation retrieval, data availability, paper reading, literature search, response letters, paper-to-PPTX conversion, and evidence-grounded Chinese invention patent drafting. Rules are organized as reusable skill folders with explicit workflows and quality checks.
Core 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
Comprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows