From claudity-assurance
Set up a new claudity-assurance QA environment — interview, scaffold directory, configure interaction surface
How this command is triggered — by the user, by Claude, or both
Slash command
/claudity-assurance:onboardThis command is limited to the following tools:
The summary Claude sees in its command listing — used to decide when to auto-load this command
<HARD-GATE> BOUNDARY: You are a BLACK-BOX QA agent. You operate ONLY within the current working directory. Do NOT read, glob, grep, or explore ANY files outside this directory — not the parent directory, not sibling directories, not the project source code. You have ZERO access to implementation details. ALL knowledge about the system under test comes from: 1. The onboarding interview with the user 2. Web search for tooling and interaction approaches 3. Direct interaction with the running system AFTER setup Do NOT "explore the project context" or "check the codebase." There is no codebase...
ALL knowledge about the system under test comes from:
Do NOT "explore the project context" or "check the codebase." There is no codebase for you. Start with the pre-flight check, then the interview.
Initialize a claudity-assurance QA environment for a new project through a collaborative interview process.
Before starting the interview:
Code detection: Check for source code markers IN THIS DIRECTORY ONLY (package.json, src/, *.csproj, go.mod, etc.). If found, STOP and instruct: "This directory contains source code. Claudity-assurance QA environments must be isolated from the codebase. Please create an empty subdirectory (e.g., qa/) and start a new session there."
Existing initialization: Check for CLAUDE.md with claudity-assurance
markers in this directory. If found, suggest /c-a:verify, /c-a:explore,
or /c-a:reset instead.
Conduct the interview one question at a time. Use multiple choice when possible.
Read the interview guide at ${CLAUDE_PLUGIN_ROOT}/skills/onboard/references/interview-guide.md
for the full question flow and adaptation patterns.
Core topics:
End the interview when there is enough to write initial docs. Do not over-interview.
Create the directory skeleton in the current directory:
CLAUDE.md — lean operating instructions with claudity-assurance markerbdds/ — dropbox for incoming BDDsdocs/ — knowledge graphresults/ — verification outputstests/ — generated test artifactschangelog.md — lightweight event logCLAUDE.md — Lean and instructional. Include:
<!-- claudity-assurance --> marker for SessionStart hook detectionKnowledge graph — Create lightweight docs in docs/ based on interview
findings. Claude decides the structure. Keep files small and linked.
changelog.md — Initialize with the onboarding event.
Based on interview:
.mcp.json in this directorytests/ or scripts/Offer to write a few initial BDD scenarios with the user as a smoke test.
Read the format spec at ${CLAUDE_PLUGIN_ROOT}/shared/bdd-format.md.
Place BDDs in bdds/.
Present what was set up and suggest next steps:
/c-a:verify when BDDs arrive from the dev session/c-a:explore to build more knowledgenpx claudepluginhub look-itsaxiom/claudity-assurance --plugin claudity-assurance/onboardGenerates four audience-tailored onboarding guides—Contributor, Staff Engineer, Executive, Product Manager—in an onboarding/ folder with index hub.
/onboardGenerates a developer onboarding guide by scanning project configs, structure, and docs for prerequisites, setup, workflow, key concepts. Writes to docs/onboarding.md.
/onboardIndexes this folder and saves a persistent overview that future sessions pick up automatically. In a git repo this writes `codebase:onboard` (code-shape map). In a non-git folder this writes `project:onboard` (file-shape index built by deterministic extractors — no LLM at index time).
/onboardGenerates a comprehensive developer onboarding guide including architecture diagram, tech stack, setup instructions, key files, API routes, database schema, and common gotchas.
/onboardGuides new Swarm users through core concepts—outcomes, modes—and launches their first team via interactive step-by-step gates.
/onboardScans new codebase, checks context and files, classifies stack, maps architecture, sets up rules, and saves knowledge to .memory/ for building.