CobraChickenAI/tml-capture-kit Skills

End-to-end architecture capture. Takes a context dump and/or conversation, produces an architecture map, inferences document, changelog, agent brief, and opportunity analysis. Orchestrates tml-extract, tml-interview, tml-infer, tml-map, tml-so-what, and tml-hygiene in one seamless experience. The user sees one flow; the skills compose underneath. This is the primary entry point — invoke this skill, not the sub-skills directly.

Extract TML primitive candidates from context dumps — documents, SOPs, wiki pages, meeting notes, or any unstructured content about a business process, team, or system. Produces structured candidate primitives for interview confirmation. Part of the TML capture pipeline — runs first, feeds into tml-interview.

Quality-control pass for TML outputs. Reviews Map, Inferences, Changelog, So What, and optional Agent Brief for consistency, provenance completeness, ambiguity, and actionability before sharing. Part of the TML capture pipeline — runs last, after tml-so-what. Final gate before sharing with stakeholders.

Surfaces inferences made during a TML interview — the gap between what was explicitly stated and what the system concluded. Produces a structured Inferences document that separates confirmed facts from model-generated conclusions, flags low-confidence inferences for human review, and makes the reasoning chain visible. Part of the TML capture pipeline — runs after tml-interview, before tml-map.

Conversational architecture extraction. Turns natural language into confirmed TML primitives through adaptive interviewing. Supports cold-start (no prior context) and warm-start (from tml-extract candidates). Produces confirmed primitives ready for tml-map. Part of the TML capture pipeline — runs after tml-extract, feeds into tml-infer.

Generate architecture artifacts from confirmed TML primitives. Produces three outputs: The Map (plain-language architecture document), The Changelog (provenance trail), and The Agent Brief (operational instructions for AI). No jargon. No TML terminology in output. Part of the TML capture pipeline — runs after tml-infer, feeds into tml-so-what.

Analyze a TML architecture map and surface specific, grounded opportunities — where AI can help, where processes can improve, where gaps create risk. The 'So What?' that turns documentation into action. Part of the TML capture pipeline — runs after tml-map, before tml-hygiene.