easydev-ai

easydev

Developer productivity toolkit for Claude Code — research, design-sync, documentation, and workflow automation.

A focused plugin with 7 essential commands that complement (not duplicate) official Anthropic plugins.

Philosophy

• Complement, don't compete — Use official plugins for code review, feature development, commits. This plugin fills the gaps.
• Smart auto-detection — Commands detect MkDocs, i18n, and project structure automatically.
• Code is truth — Every command treats code as the source of truth, not assumptions.
• Parallel by default — Research and auditing spawn multiple sub-agents for speed.

Installation

From Marketplace

# Add the easydev-ai marketplace
/plugin marketplace add easydev-ai/easydev

# Install the plugin
/plugin install easydev@easydev-ai

Manual Installation

# Clone to your plugins directory
git clone https://github.com/easydev-ai/easydev ~/.claude/plugins/easydev

Restart Claude Code after installation.

Commands

Command	Description	Use When
/easydev:research	Deep research + codebase investigation with parallel sub-agents	"Is X applicable to us?" "Why is Y breaking?"
/easydev:design-sync	Bidirectional code ↔ design doc alignment	Code and spec have drifted apart
/easydev:docs-audit	Audit docs for duplicates, orphans, broken links (auto-detects MkDocs)	Documentation cleanup needed
/easydev:docs-refresh	Batch-update docs to match current code state	Code changed, docs are stale
/easydev:synthesize	Distill conversation into structured documentation	Long discussion needs to become permanent docs
/easydev:standup	Generate standup notes from git activity	Daily standup prep
/easydev:onboard	Generate comprehensive onboarding documentation	New team member joining

Command Details

/easydev:research <question> [code-paths...]

Your most powerful research tool. Spawns parallel sub-agents to investigate:

# Research a technology
/easydev:research Is API Gateway HTTP API applicable to our lambda/processor?

# Investigate a bug
/easydev:research Why are transcriptions failing for large files? src/services/audio/ --mode investigate

# Compare options
/easydev:research Should we switch from REST to GraphQL?

Modes:

• research — External research, technology evaluation
• investigate — Deep codebase analysis, dependency mapping
• auto (default) — Intelligently combines both

Features:

• Parallel sub-agent execution (4+ agents at once)
• Blast radius analysis for code changes
• Evidence-based recommendations with sources

---

/easydev:design-sync <design-doc-path> [code-path]

Identifies mismatches between design documents and code, then asks YOU which direction to fix:

/easydev:design-sync docs/specs/auth.md src/auth/

Output:

| # | Design Doc Says | Code Does | Which is Correct? |
|---|-----------------|-----------|-------------------|
| 1 | Password reset required | Not implemented | ? |
| 2 | localStorage tokens | httpOnly cookies | ? |

Respond: "1: doc, 2: code" to specify fix direction

Why this is unique: Most tools assume the spec is always right. This command recognizes that sometimes code evolved past the spec.

---

/easydev:docs-audit [target-path]

Smart documentation auditing that auto-detects your documentation system:

# Audit all docs
/easydev:docs-audit docs/

# Focus on specific issues
/easydev:docs-audit --focus duplicates
/easydev:docs-audit --focus translations --lang zh

Auto-detects:

• MkDocs (checks for mkdocs.yml) → i18n-aware, nav validation
• Generic → File-based auditing, duplicate detection

Finds:

• Duplicate content (exact and semantic)
• Orphan files (not linked anywhere)
• Broken internal links
• Translation gaps (MkDocs mode)
• Nav mismatches (MkDocs mode)

---

/easydev:docs-refresh [docs-path] [code-path]

Batch-update documentation to match the current code state:

# Scan and update all docs
/easydev:docs-refresh

# Focus on specific areas
/easydev:docs-refresh docs/api src/routes --scope critical

# Scan only (no changes)
/easydev:docs-refresh --mode scan

Modes:

• scan — Report staleness only, don't propose changes
• update — Propose and apply changes with approval
• auto (default) — Scan first, then offer to update

Scopes:

• all — Check every doc against code
• stale (default) — Only docs that appear outdated
• critical — Only docs with breaking inaccuracies

How it works:

1. Maps documentation structure (auto-detects MkDocs)
2. Maps code exports, routes, configs, env vars
3. Spawns parallel agents to compare docs vs code
4. Reports staleness with severity (🔴 Critical, 🟡 Stale, 🟢 Fresh)
5. Proposes surgical updates (preserves human context)
6. Applies changes only with your approval