orch-add-feature

SkillAI & models

orch-add-feature is a skill that orchestrates building a brand-new feature end to end. It classifies the feature's size, then delegates each phase, research, planning, test-first implementation, review, and a gated commit, to the matching ECC agent. It is meant for adding a capability that does not exist yet, not for fixing defects or changing existing behavior.

Available today. Use it from your connected AI after setup.

Have the ECC agents available so the skill can delegate each phase to them.

Then ask your AI: use the orch-add-feature skill

What your AI can do with it

  • Classifies a feature's size and right-sizes the workflow accordingly
  • Delegates research of existing solutions to a specialist agent
  • Produces a plan and stops at Gate 1 for plan approval
  • Implements test-first: writes new failing tests, then code to green
  • Runs code review, adding security-reviewer when a security trigger is touched
  • Stops at Gate 2 for confirmation before committing

Getting started

  1. Have the ECC agents available so the skill can delegate each phase to them.
  2. Have the related orch-pipeline skill in place, since this skill is a thin wrapper over its shared engine.
  3. Ask your agent to add a capability that does not exist yet, naming the feature you want.
  4. Approve the plan at Gate 1 before implementation begins.
  5. Confirm the commit at Gate 2 after review completes.

What this skill tells your AI

The instructions your AI receives, as published by affaan-m/ecc in skills/orch-add-feature/SKILL.md and read by ahel’s review.

Actor · action · target: orch · add · feature. Thin wrapper over the shared engine in orch-pipeline.

When to Use

  • The user wants a capability that does not exist yet ("add", "build", "implement", "support …").
  • It is net-new behavior — not a correction (orch-fix-defect) and not an alteration of existing behavior (orch-change-feature).

Operation settings

  • Default size floor: standard — run Research + Plan unless clearly small.
  • Phase mask: 0 → 1 → 2 → 4 → 5 → 6 (skip 3 Scaffold; that is MVP-only).
  • First move (phase 4): write new failing tests for the new behavior, then implement to green.

How It Works

  1. Run the orch-pipeline engine with the settings above.
  2. Classify size first; small / trivial features collapse toward 4 → 5 → 6.
  3. Stop at Gate 1 (plan approval) and Gate 2 (pre-commit).
  4. Add security-reviewer if the feature touches a security trigger.

Related: /feature-dev is a standalone version of this flow. orch-add-feature differs by sharing the orch-pipeline engine — the size classifier and the two gates — with the rest of the family, so it right-sizes trivial features to 4 → 5 → 6.

Example

orch-add-feature: add OAuth2 login to nws-poller
→ research existing auth libs → plan task_list  [GATE 1: approve]
→ TDD each task → code-review (+ security-reviewer: auth path)
→ commit  [GATE 2: confirm]

Signals

GitHub stars
268k
Forks
40k
Last commit
Sep 2026

Questions

How do I manually add a skill to Claude?
A skill is a file with a name and description, like the orch-add-feature skill shown here, placed where your agent can load it. The exact install location depends on the app you use, so follow that app's skill documentation.
What are some good skills to add to Claude?
Skills that match work you repeat, such as orchestrating a new feature end to end, fixing defects, or changing existing behavior. This item covers net-new features; related skills handle corrections and alterations.
Advanced
Item type
skill
Key
orch-add-feature
Source
github.com/affaan-m/ecc