By ericmittman
Opinionated engineering workflow: superpowers chained end-to-end — skill routing nudges, execution-strategy analysis, automatic simplification passes, and a /feature pipeline command
Stage 2 of the engineering-flow pipeline — Opus-pinned plan writing, security-annotated, with managed (optionally nested-parallel) reviewed execution.
Stage 3 of the engineering-flow pipeline — final adversarial assessment (spec coverage, integration seams, full security+reliability review). Run on your strongest model. Evaluates only; builds nothing.
Stage 1 of the engineering-flow pipeline — security-aware brainstorm + spec. Run on your strongest model; hand off to /feature-build.
Durable build state for /feature-build — journal format, update cadence, and resume protocol. The journal is what lets a build survive context loss and what /feature-review assesses against.
Evaluate a plan's task dependencies to determine serial vs parallel execution — builds a dependency graph, identifies the critical path, and recommends the right approach without asking
Security AND reliability posture for the engineering-flow pipeline — design-time questions, [SEC]/[REL] task-flagging criteria, and review rubrics. One source of truth consumed by /feature, /feature-build, and /feature-review.
Stack-agnostic code simplification triggered by the smart-simplify Stop hook after substantial changes
Operating contract for a nested track-coordinator subagent in /feature-build parallel mode — scope discipline, internal review loop, escalation, and report format. Requires Claude Code >= 2.1.172 (nested subagents).
Executes bash commands
Hook triggers when Bash tool is used
Modifies files
Hook triggers on file write and edit operations
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
No model invocation
Executes directly as bash, bypassing the AI model
No model invocation
Executes directly as bash, bypassing the AI model
An opinionated engineering workflow for Claude Code: the superpowers discipline chained end-to-end, plus the connective tissue that makes it run itself.
/feature (your best model) ─▶ security-aware brainstorm ─▶ spec
└▶ /feature-build (opus) ─▶ [SEC]/[REL]-annotated plan ─▶ execution-strategy
─▶ reviewed execution (sonnet builds · opus reviews · haiku verifies
· nested parallel tracks on CLI ≥ 2.1.172, flat otherwise)
└▶ /feature-review (your best model) ─▶ coverage + seams + full security audit
<idea> — one command runs the whole pipeline, honoring every gate.~/.claude/skill-routes.local.conf.One rubric file (secure-design skill) drives all three stages: design-time questions
in the brainstorm, [SEC]/[REL] flags on plan tasks, inline diff checks on flagged
tasks during the build, and an unconditional full-diff security + reliability
assessment at review. The reliability half exists because untimeouted blocking waits
are the most common serious bug we ship — the rubric makes them un-shippable quietly.
Design and final review run on your session's strongest model. /feature-build pins
itself to Opus. Inside the build: Sonnet implements, Opus reviews, Haiku runs
verification commands and inventories (never reviews, never decides). No Fable/Opus
access? Stages degrade to your session model — the pipeline still works.
git clone https://github.com/ericmittman/engineering-flow && cd engineering-flow
./install.sh # add --dry-run to preview
Or manually, inside any Claude Code session:
/plugin install superpowers@claude-plugins-official
/plugin marketplace add ericmittman/engineering-flow
/plugin install engineering-flow@engineering-flow
/feature add a healthcheck endpoint →
brainstorming announces and asks a question (no code); approve the spec, then
/feature-build writes a flagged plan before any implementation.npm test (or pytest/go test) → the router nudges test-driven-development./plugin → remove engineering-flow (and superpowers if you want). Nothing else to
clean — this kit never touches your settings.json.
Add lines to ~/.claude/skill-routes.local.conf (created by you, survives kit
updates): pattern | skill | path | description, where pattern is a grep -Ei regex
matched against your Bash commands. Route to any skill you have installed.
If you previously ran these hooks from ~/.claude/settings.json directly: remove the
personal skill-router/accumulate-simplify/smart-simplify-stop entries after installing
the kit (one router, one simplify pair — never both), and migrate personal routes to
~/.claude/skill-routes.local.conf.
npx claudepluginhub ericmittman/engineering-flow --plugin engineering-flowHarness-native ECC operator layer - 67 agents, 271 skills, 92 legacy command shims, reusable hooks, rules, selective install profiles, and production-ready workflows for Claude Code, Codex, OpenCode, Cursor, and related agent harnesses
Superpowers Plus core skills library for Claude Code: planning, execution routing, TDD, debugging, and collaboration workflows
Core skills library for Claude Code: TDD, debugging, collaboration patterns, and proven techniques
Persistent file-based planning for AI coding agents. Crash-proof markdown plans (task_plan.md, findings.md, progress.md) that survive context loss and /clear, with an opt-in completion gate and multi-agent shared state. Manus-style. Works with Claude Code, Codex CLI, Cursor, Kiro, OpenCode and 60+ agents via the SKILL.md standard. Includes Arabic, German, Spanish, and Chinese (Simplified and Traditional).
Unity Development Toolkit - Expert agents for scripting/refactoring/optimization, script templates, and Agent Skills for Unity C# development
Claude harness - A harness for solo developers (Vibecoders) to handle full-cycle contract development.