Skip to main content
Reference — generated from the toolkit

Source of truth: toolkit/README.md. Edit it there; this page is regenerated on build.

The RAPID toolkit

Machine-consumable artefacts for the RAPID process — the things you instantiate or an agent loads verbatim. The written handbook (process definition, standards, playbooks, gate rationale) lives in the Docusaurus site and is the human reading source; this tree holds the operational counterparts those pages describe.

DirectoryContentsStatus
gates/Operational gate cards — the checklist a lead or agent runs at each phase transitionv1 (8 cards + template)
templates/Fillable artefact templates — ADR, PFD, domain brief, UX flow, data model, screen spec, runbook, risk-acceptance, release-evidence, incident-comms, test-plan, scorecard, squad-charter, stack-playbook, legacy-knowledge-pack, competitive-teardown, demo-scriptv1 (17)
scaffold/The RAPID project scaffold — the skeleton every build starts from (agent context, memory, ADR/runbook dirs, CI)v1 skeleton
skills/Agent skills — spec + golden tests per skill. Specs: code-archaeology, characterisation-testsv1 (2 specs + template)
radar/Tech radar as code (rings: adopt / trial / assess / hold) — radar.json seededv1 seed
prompts/The versioned prompt library — one prompt per file with front-matter. Seeded across definition / engineering / test / security / design / remediation / gtmv1 (7 seeded)
registers/Living registers — tech-debt, security-debt, EOL inventory, security exceptions, client & market signals (schemas)v1 (5 schemas)
metrics/Metrics framework definitions + platform configscaffold
training/Role-based curriculum modulesscaffold

Everything here is versioned, owned, and written to be read by humans and loaded by agents. If an agent can't be pointed at an artefact, it doesn't exist.