Drives the Compound Engineering commands as one sequence. On first use in a repo it readies the project automatically and without blocking — installs compound-engineering if missing (asks first) and sizes up the stack — then routes feature vs bug, runs the /ce-* steps in order, gates each on a real artifact, and stops only to approve the plan and the result. Requires nothing of a repo that CE does not. Independent companion; not affiliated with Every.
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.
ce-conductor runs the Compound Engineering commands from the Compound Engineering (CE) Claude Code plugin (repo, site) as one workflow. It checks that each step worked before starting the next, and it remembers where it is.
CE gives you a set of commands for building software with AI. You normally run them one at a time, in order, and keep track of where you are yourself.
ce-conductor adds three things on top of CE:
.flow/state.md, so it can resume a half-finished run and always show you the next step.It stops to ask you at only the two points that need a decision:
It works with any codebase that CE works with, and it handles the most common kinds of work: new features, bug fixes, and small changes.
This plugin is a thin add-on. It does not replace CE or change how CE works. It only runs CE's commands, in order, with the checks and the memory above. It is not affiliated with Every or its maintainer.
CE is a separate Claude Code plugin, made by Every, that breaks development into steps you run as commands:
ce-brainstorm: turn a rough idea into a short requirements documentce-plan: turn that document into an implementation plance-work: carry out the plan and write testsce-code-review: review the changes before they mergece-compound: record what was learned so the next task is easierce-debug: reproduce and fix a specific bugce-ideate: weigh several bigger ideas before committing to onece-strategy: keep a short STRATEGY.md that the planning steps read as backgroundce-conductor needs CE installed because these are the commands it runs.
ce-conductor sorts each request into one of three sequences. The feature and bug sequences are CE's own documented examples. The small-change sequence is a trimmed feature path.
ce-debug, then ce-code-review, then ce-compound. The fix is not treated as done until a regression test reproduces the original failure and passes after the fix.ce-brainstorm, then ce-plan, ce-work, ce-code-review, and ce-compound. For a large or unclear feature it can start with ce-ideate to weigh options first.ce-plan, then ce-work, ce-code-review, and ce-compound. For obvious, low-risk changes that need no brainstorming.These cover the common kinds of work. Refactors, chores, and investigations run as a small change unless their scope calls for the feature sequence. If a STRATEGY.md file exists, CE reads it as background during brainstorm and plan, so you do not run a strategy step yourself.
Compound Engineering is required. You do not install it separately. If missing, ce-conductor installs it the first time you run /ce-conductor and asks before doing so. If you already have CE, ce-conductor uses it.
Run both commands in Claude Code. The first registers where the plugin lives. The second installs it.
/plugin marketplace add robertcdawson/ce-conductor
/plugin install ce-conductor@ce-conductor
Restart Claude Code. You will have one command, /ce-conductor. Run /help to confirm the exact name if your version lists it with a prefix.
Give the conductor a feature or a bug:
/ce-conductor "add a --version flag to the CLI"
/ce-conductor "checkout webhook sometimes double-charges"
The first time you run it in a repository, it checks that CE is installed and looks at what test, lint, and build commands the project has, so its checks have something real to verify against. This asks nothing of you or the repo. If the project has no tests, it continues anyway and tells you that verification is limited. After the first run it skips this check.
It then runs the matching CE sequence, writes its progress to .flow/state.md, and pauses only to let you approve the plan and approve the result.
/ce-conductor "<request>"npx claudepluginhub robertcdawson/ce-conductor --plugin ce-conductorComprehensive feature development workflow with specialized agents for codebase exploration, architecture design, and quality review
Harness-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
Continuous self-referential AI loops for interactive iterative development, implementing the Ralph Wiggum technique. Run Claude in a while-true loop with the same prompt until task completion.
Feature development with code-architect/explorer/reviewer agents, CLAUDE.md audit and session learnings, and Agent Skills creation with eval benchmarking from Anthropic.
Real-time statusline HUD for Claude Code - context health, tool activity, agent tracking, and todo progress
Comprehensive skill pack with 66 specialized skills for full-stack developers: 12 language experts (Python, TypeScript, Go, Rust, C++, Swift, Kotlin, C#, PHP, Java, SQL, JavaScript), 10 backend frameworks, 6 frontend/mobile, plus infrastructure, DevOps, security, and testing. Features progressive disclosure architecture for 50% faster loading.