Convert vague software requests into deterministic projects via Markdown knowledge graphs.
npx skills add https://github.com/RazvanBd/ai-composer --skill ai-composer
Please help me install this Agent Skill. Skill: ai-composer Source: https://github.com/RazvanBd/ai-composer/tree/main Command: npx skills add https://github.com/RazvanBd/ai-composer --skill ai-composer
The AI Composer skill enables deterministic orchestration of software projects by coordinating specialized AI agents and a disk-based knowledge graph built from Markdown artifacts.
Initialize the knowledge graph from Markdown artifacts and run the orchestrator to generate per-ticket contexts.
High-intent search queries and answers about installing and using this skill.
Deterministic software project generation is achieved by orchestrating specialized AI agents using a disk-based knowledge graph of Markdown artifacts. The system coordinates roles like Architect and Coder to translate vague requests into structured, typed outputs.
A Markdown-backed knowledge graph is a disk-stored structure of Epics, Tickets, ADRs, and Business Rules used to guide AI orchestration. It provides the context required to generate strongly-typed TicketContexts for software delivery.
Yes, AI orchestration converts vague development requests into deterministic projects by generating a strongly-typed TicketContext per ticket. It applies this context to a Markdown knowledge graph to drive provider-driven agent execution.
State and telemetry for AI-generated software artifacts are persisted on disk alongside Markdown-backed project files. This ensures auditability and resilience across the lifecycle of each ticket context generated by the orchestrator.
Yes, provider-driven agent execution supports sandboxed code generation and safe workspace file application with validation. This ensures that generated outputs are applied securely to the target workspace without risking unvalidated modifications.
Automating software project delivery requires orchestrating specialized AI agents including PO, PM, BA, Software Architect, Tech Lead, Coder, Reviewer, and QA. These roles collaborate to build the knowledge graph and generate per-ticket outputs.
humansysDesign epic scope, stories, and architecture for multi-story work.
agentclashManage E2B-based coding agent evaluation harnesses and execution workflows.
linenoizeSave coding session checkpoints, learnings, and project state for AI assistants.
erclxManage markdown prompt snippets across projects with the aitk CLI.
