ub

Plugin ub from marketplace ub-marketplace in BilLogic/uno-blueprint. Commands read from the repo's manifest — 12 days ago.

Skills (4)

• audit

  Blueprint audit — run the consistency-check roster over an imported blueprint (gaps, jargon, channel conflicts, KPI drift, ownership, value dead-ends, hidden fees), recording findings ranked by impact and effort; re-run the checks, or dismiss or resolve a finding. Building → ub:map; hypothetical cha

• map

  Service blueprints — create one from service docs or discovery research, mapping how the service actually runs; import or translate a FigJam, Figma, Miro or spreadsheet blueprint; present, update or redeploy a blueprint workspace (a directory holding blueprint-workspace.json or blueprint/). Journey

• slice

  Blueprint slice — cut a stakeholder view out of an imported, signed-off blueprint: one actor's journey, one moment across every lane, one lane end to end (what a team owns), one cell brief, or a slide-by-slide walkthrough, text or illustrated, for a client or exec; writes it back as a slice the app

• whatif

  Blueprint what-if — trace a hypothetical change to an imported blueprint before anyone commits: replay a scenario as if it were made (what breaks, where displaced load lands), restage a moment across the visibility line, or prioritize which cells to focus on next; promote an accepted analysis into t

Subagents (5)

• auditor

  Dispatched by ub:audit, one per roster check, in parallel: runs exactly one check against a blueprint export and returns findings JSON. Blind by design — it sees its one check doc and the export alone, so each judgement stays uncontaminated. Read-only: it returns findings to the caller.

• blueprint-reviewer

  Dispatched after the validator exits 0 and before the write: a fresh-context adversarial review of a drafted blueprint IR before sign-off, of a slice file before import (slice mode), or of a whatif analysis (whatif-claim mode). Returns numbered findings with severities — referential gaps, journey-lo

• document-reader

  Dispatched in parallel fan-outs during ingestion or translation, piloted on a few documents first: reads source documents and returns their structure, keeping raw text out of the main conversation. Three modes, set by the dispatching prompt — corpus survey (journey / system / sensitive-exclude), sin

• impact-tracer

  Dispatched by ub:whatif, and by map's update mode for chain questions: walks the dependency graph from a proposed change or a named cell along `leads_to` and `enables` edges, downstream by default or upstream, and returns the affected cells, the assumptions the change breaks, and where displaced dem

• render-checker

  Dispatched after an import's read-back verification passes, against a local dev server or a deployed URL: drives a browser through every scenario and layout (stacked and merged, every path), screenshots each, collects console errors, and returns a pass/fail report per scenario. Runs with the full to

Hooks (3)

• SessionStart

  bash "${CLAUDE_PLUGIN_ROOT}/hooks/session_start_status.sh"

• PostToolUse

  python3 "${CLAUDE_PLUGIN_ROOT}/hooks/validate_ir_on_edit.py"

• PreToolUse

  python3 "${CLAUDE_PLUGIN_ROOT}/hooks/secret_guard.py"

Similar plugins