Run spec-first agentic engineering workflows with Flow-Next
Turn rough coding intent into durable specs, task graphs, fresh-context worker runs, adversarial reviews, PR bodies, and evidence receipts.
npx skills add agentskillexchange/skills --skill run-spec-first-agentic-engineering-workflows-with-flow-next
Use Flow-Next when a coding-agent operator needs a full lifecycle workflow from idea capture through planning, implementation, review, PR creation, and feedback resolution. The workflow creates repository-local .flow/ specs and task records, runs named /flow-next:* skills, re-anchors workers against durable context, and records commits, test evidence, review verdicts, and receipts. This is not a hosted project-management product or generic orchestration framework: the boundary is repo-local, spec-first agentic engineering across Claude Code, Codex, Cursor, Droid, and compatible agent hosts.
What this skill actually does
Inputs and prerequisites: Claude Code, Codex, Cursor, Factory Droid, or compatible agent host; git; jq and gh for review and PR workflows.
Setup notes: Install through the host-specific plugin path, such as /plugin install flow-next in Claude Code or ./scripts/install-codex.sh flow-next for Codex, then run /flow-next:setup inside each target repository.
Source and verification boundary: use https://flow-next.dev as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a Multi-Framework workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.