vt-c-dev-start¶

Developer entry point — shows which of the toolkit's many skills matter for your task (feature, bug, UI, prototype, skill-building, or debugging). Use this when you are a developer who wants a short, task-based map instead of scanning the full skill list.

Plugin: core-standards
Category: Other
Command: /vt-c-dev-start

Dev Start — Task-Based Skill Map¶

You are a developer (Dominik, Markus, Niclas-on-a-dev-task, or anyone writing code with this toolkit). The toolkit ships well over a hundred skills — far more than anyone can remember. This skill answers one question: "For the task in front of me, which handful of skills should I reach for, and where do I start?"

It does NOT run a workflow. It prints a map and points you at the right skill.

Disambiguation — three "start" skills¶

Skill	Answers
`/vt-c-dev-start` (this)	"I'm a dev — which skills apply to my task?"
`/vt-c-0-start`	"How does the VT development flow run (phases 0–6)?"
`/vt-c-0-init`	"How do I start a brand-new project?"

If the user wanted the phase workflow overview, redirect them to /vt-c-0-start. If they wanted to scaffold a new project, redirect to /vt-c-0-init.

Execution¶

Step 1 — Resolve the task category¶

If it matches a category → show that category's map (Step 3).
If it is show-all, empty, or unrecognized → show all six category headers with their skills (Step 3 for every category) and this pointer: "For a role-level overview, read docs/getting-started/by-role.md; for the full task view, docs/getting-started/by-task.md."

Step 2 — Load authoritative skill metadata¶

The skill lists below are curated names. For the authoritative one-line description of any skill, read it from docs/skill-orchestration-map.json (the SPEC-125 orchestration map: skills[].name → skills[].description). Do not paraphrase from memory — read the JSON so descriptions never drift. If a listed skill is absent from the JSON (renamed/removed), say so and point the user to docs/getting-started/by-role.md rather than inventing a description.

Step 3 — Render the map for the resolved category¶

Print the category header and, for each skill, name — <one-line context>. Pull the full description from the JSON (Step 2) when the user wants more.

feature — build a feature end-to-end - /vt-c-activate — load a spec (or generate one from a PRD) and start the cycle - /vt-c-2-plan — research + architecture + task breakdown before you code - /vt-c-3-build — implement against the plan with decision journaling - /vt-c-4-review — mandatory multi-agent quality + security review - /vt-c-test-driven-development — RED→GREEN→refactor for the code you write

bug — fix a reported bug - /vt-c-bug-report — file a tracked BUG-NNN intake so the fix is spec-backed - /vt-c-triage-bugs — promote / defer / dedupe pending bug reports - /vt-c-systematic-debugging — structured hypothesis-driven debugging - /vt-c-root-cause-tracing — trace an error backward through the call stack - /vt-c-continuous-learning — feed the fix back so the mistake is not repeated

ui — build or adjust user interface - /vt-c-frontend-design — distinctive, production-grade frontend code - /vt-c-visitrans-design-system — apply VisiTrans Corporate Design tokens - /vt-c-pd-3-prototype — generate a deployable Angular + PrimeNG prototype - /vt-c-c4-diagram — C4 context/container diagrams for the UI's system view - /vt-c-mermaid-diagrams-branded — branded Mermaid diagrams for flows/screens

prototype — spin up a throwaway / POC - /vt-c-scaffold — universal folder structure + CLAUDE.md + vision/PRD stubs - /vt-c-bootstrap — best-practice project structure, CI, docs folders - /vt-c-kw-prototype — deployable Angular prototype with mock data - /vt-c-frontend-design — quick, polished UI when the POC is frontend-facing - /vt-c-pd-3-prototype — CD-compliant prototype when it will inform a handoff

skill — build a skill, agent, or MCP server - /vt-c-skill-creator — guided creation of a new skill - /vt-c-create-agent-skills — author/refine Claude Code skills (SKILL.md) - /vt-c-mcp-builder — build a Model Context Protocol server - /vt-c-agent-native-architecture — prompt-native agent design patterns - /vt-c-skill-eval — run eval cases against a skill before you ship it

debug — deep / systematic debugging - /vt-c-systematic-debugging — structured root-cause workflow - /vt-c-root-cause-tracing — backward trace from a deep failure - /vt-c-defense-in-depth — validate at every layer so bugs become impossible - /vt-c-verification-before-completion — prove "it works" before claiming done - /vt-c-recursive-criticism — self-review pass that catches ~40% more issues

Step 4 — Offer the next step¶

After the map, offer to run the most likely first skill for the category, e.g. for feature: "Want me to start with /vt-c-activate?" — but do not run it without the user's go-ahead.

Notes¶

This map is a curated subset (~5 skills × 6 tasks). It is deliberately short — the full inventory lives in docs/skill-orchestration-map.md and the deeper, role-oriented view in docs/getting-started/by-role.md.
Skill descriptions are sourced from docs/skill-orchestration-map.json (regenerated by scripts/generate-skill-map.sh) so they stay in sync with the deployed skills and never drift from a hand-maintained copy.