TML Capture Kit

Capture architecture from conversation. One flow, seven skills built on the open Agent Skills standard, a plain-language map anyone can read.

What this does

You give it documents about how something works -- SOPs, wiki pages, meeting notes, whatever you have. It asks a few questions. You get back a complete architecture map with visible reasoning and actionable opportunities.

Total time for the human: about 20 minutes.

The pipeline

CONTEXT    ──▶  INFORMED    ──▶  SURFACE     ──▶  THREE      ──▶  "SO       ──▶  QUALITY
DUMP            INTERVIEW        INFERENCES       ARTIFACTS       WHAT?"         GATE

tml-extract     tml-interview    tml-infer        tml-map         tml-so-what    tml-hygiene
10 min          10 min           auto             auto            auto           auto

Stage	Skill	What it does
1	tml-extract	Pulls candidate primitives from unstructured documents
2	tml-interview	Confirms and refines candidates through conversation
3	tml-infer	Separates what was said from what was concluded
4	tml-map	Generates the Map, Changelog, and Agent Brief
5	tml-so-what	Surfaces opportunities, friction, and quick wins
6	tml-hygiene	Quality gate before sharing with stakeholders

tml-capture is the orchestrator -- it runs the full pipeline as one seamless experience. Invoke it, not the sub-skills.

What you get back

Artifact	What it is	Who it's for
The Map	Plain-language architecture -- what you do, who does it, what the rules are	Everyone
The Changelog	Where every element came from	Anyone asking "where did this come from?"
The Inferences	What the system concluded vs. what was stated	The person auditing the map
The Agent Brief	Operational instructions for AI agents	Any AI working within this scope
The So What?	Opportunities, friction points, quick wins	The person deciding what to do next
Hygiene Report	Quality check -- consistency, provenance, actionability	The person before they hit send

Install

These skills follow the open Agent Skills specification and work with any compatible agent.

Each skill directory is self-contained — SKILL.md plus its own references/ with output schemas and templates.

Claude Code

git clone https://github.com/CobraChickenAI/tml-capture-kit.git
cp -r tml-capture-kit/skills/tml-* /path/to/your-project/.claude/skills/

Codex

npx skills add CobraChickenAI/tml-capture-kit

Cursor / VS Code

git clone https://github.com/CobraChickenAI/tml-capture-kit.git .cursor/skills/tml-capture-kit

Or download the repo as a zip and extract the skills/ directory into your project.

Data contracts

Every handoff between skills has a formal schema. Each skill carries the schemas for its own outputs in its references/ directory. The orchestrator (tml-capture) holds all schemas and the pipeline definition (pipeline.yaml) — the machine-readable DAG that governs execution order, data flow, and human gates.

Part of The Missing Layer

This kit is part of the TML ecosystem. TML provides the primitives (Scope, Domain, Capability, Archetype, Policy, Connector, Binding, View, Provenance) that give the capture pipeline its structure.

License

MIT. See LICENSE.