Infer Garura product LTM artifact proposals from codebases via deterministic scanning.
npx skills add https://github.com/kapilvirenahuja/garura --skill garura-codify
Please help me install this Agent Skill. Skill: garura:codify Source: https://github.com/kapilvirenahuja/garura/tree/main/core/components/plays/codify Command: npx skills add https://github.com/kapilvirenahuja/garura --skill garura-codify
Codify eliminates the costly brownfield gap between an existing codebase and the structured Garura product LTM it already embodies, turning code reality into a complete, validator-compatible proposal tree without requiring user interviews.
{stm_base}/{issue}/evidence/codify/, with zero writes to product LTM; promotion to LTM happens only after the user Tethers and later via /garura:enrich.Use the codify skill to infer your Garura product LTM proposals from the current repository by staging outputs under your STM evidence paths.
High-intent search queries and answers about installing and using this skill.
You can infer product architecture from an existing codebase by using deterministic code scanning to generate a scan index, followed by agent-based inference to construct foundation, scope, epic, and architecture proposals.
Bootstrapping a brownfield migration for multi-repo projects requires executing a full pipeline orchestration that scans the codebase and infers a complete artifact tree without requiring user interviews.
Schema validation for inferred architecture proposals works by validating generated outputs against target schemas using JSON contracts, ensuring all inferred artifacts are structurally sound before committing evidence.
Yes, you can run codebase analysis safely by staging all inferred proposals as evidence in short-term memory only, ensuring zero writes to product long-term memory until an explicit tether is triggered.
No, intent-driven development for existing repositories does not require manual mapping because agentic orchestration automatically reverses output shapes from deterministic scans to infer the original design intent.
A limitation of using deterministic scanning for architecture inference is that it requires a strictly defined workflow to generate a scan index first, and all inferred output proposals remain staged until a manual checkpoint gate is passed.
kapilvirenahujaScan repository roots to generate scan-index.json and scan-result.yaml artifacts.
kapilvirenahujaAnalyze merged PR diffs and issue STM evidence to generate up to two staged proposals.yaml entries.
photon-groveOrganize session artifacts and layered memory files in repository workflows.
samdaeGenerate spec and architecture markdown files from backend and frontend code.