scout-plugins

name: scout-plugins description: | Research new/updated Claude Code plugins, MCP servers, and skills released in the last week (or custom time window). Evaluates each against the current dev-team plugin setup, identifies gaps, and proposes enhancements with priority and effort. Usage: /scout-plugins or /scout-plugins 2w or /scout-plugins 30d

argument-hint: "[time window, e.g. 1w, 2w, 30d — default: 1w]"

Scout Plugins

Systematic competitive intelligence scan of the Claude Code plugin ecosystem.

Arguments

• /scout-plugins — scan the last 1 week (default)
• /scout-plugins 2w — scan the last 2 weeks
• /scout-plugins 30d — scan the last 30 days
• /scout-plugins 3m — scan the last 3 months

Parse the argument to determine the time window. Default to 1 week if omitted.

Step 0: Load current capabilities

Read in parallel to understand what the dev-team plugin already provides:

_gc=$(git rev-parse --git-common-dir 2>/dev/null) \
  && MROOT=$(cd "$(dirname "$_gc")" && pwd) \
  || MROOT=$(pwd)

• $MROOT/README.md — current feature set, commands, agents (changelog lives in $MROOT/CHANGELOG.md)
• $MROOT/AGENTS.md — project rules and conventions
• $MROOT/.claude-plugin/plugin.json — current version

Build a mental inventory of:

• All commands/skills and what they do
• All agents and their capabilities
• Review workflow, memory system, TDD gates, orchestration features
• Recent changelog entries (what was added recently — avoid re-discovering it)

Step 1: Research (parallel web searches)

Run these searches in parallel, adjusting date terms for the time window:

Search batch 1 (plugin ecosystem)

1. Claude Code plugins released updated <month> <year>
2. Claude Code MCP servers new plugins <month> <year>
3. Claude Code plugin marketplace new releases <date range>

Search batch 2 (community & repos)

4. site:github.com claude code plugin <year> <month>
5. awesome claude code extensions plugins <year>
6. Claude Code skills hooks agents new <month> <year>

Search batch 3 (specific categories)

7. Claude Code code review plugin <year>
8. Claude Code memory context plugin <year>
9. Claude Code testing TDD plugin <year>
10. Claude Code orchestration workflow plugin <year>

Collect all unique plugins, skills, MCP servers, and tools mentioned.

Step 2: Filter and deduplicate

From the search results, build a candidate list. For each candidate:

1. Name and author
2. What it does (1-2 sentences)
3. When released/updated (within the time window?)
4. Install count or stars (if available)
5. Source URL (GitHub repo or marketplace link)

Discard:

• Anything released before the time window
• Anything already installed or incorporated into dev-team (check changelog)
• Anything clearly low-quality (no README, no stars, abandoned)

Step 3: Deep evaluation of top candidates

For each candidate that passes the filter (up to 10), do a deeper dive:

3a. Fetch details

Read the plugin's README, SKILL.md, or documentation page. Understand:

• Exact features and commands provided
• Architecture (sub-agents, hooks, MCP servers, etc.)
• Dependencies and requirements
• License

3b. Gap analysis against dev-team

For each candidate, answer:

3c. Classify

Assign each candidate to one of:

• ADOPT — install or incorporate this; clear value add
• STEAL — don't adopt wholesale, but steal specific ideas/patterns
• WATCH — interesting but not actionable yet; revisit next scan
• SKIP — not relevant or already covered

Step 4: Output the report

═══ PLUGIN SCOUT REPORT ═══════════════════════════════
Time window: <start date> → <end date>
Current version: dev-team v<version>
Candidates scanned: <N>
═══════════════════════════════════════════════════════

## ADOPT (clear value — incorporate into dev-team)

### <plugin-name> by <author> — <stars/installs>
What: <1-2 sentences>
Gap: <what dev-team is missing>
Effort: <low/medium/high>
Source: <URL>

---

## STEAL (borrow ideas, not the whole plugin)

### <plugin-name> by <author>
What: <1-2 sentences>
Idea to steal: <specific technique or pattern>
How to apply: <where in dev-team this would go>
Effort: <low/medium/high>

---

## WATCH (revisit next scan)

- <plugin-name> — <why it's interesting but not ready>

---

## SKIP (already covered or not relevant)

- <plugin-name> — <reason>

═══════════════════════════════════════════════════════

Omit any section with no entries.

Step 5: Enhancement proposal

If any ADOPT or STEAL candidates were found, produce an enhancement table:

## Proposed Enhancements

| Priority | Enhancement | Inspired By | Effort | Affects |
|----------|-------------|-------------|--------|---------|
| High     | <what>      | <plugin>    | Low    | <files> |
| Medium   | <what>      | <plugin>    | Medium | <files> |

Step 6: Save the report

Save the full report to:

WTROOT=$(git rev-parse --show-toplevel 2>/dev/null || pwd)
# .claude/plans/<YYYY-MM-DD>-scout-plugins.md

Print:

Report saved to: .claude/plans/<date>-scout-plugins.md

Next steps:
  - Review ADOPT items and decide which to implement
  - Schedule implementation with /kickoff for each approved enhancement
  - Run /scout-plugins again in <time window> to stay current

Rules

• Always compare against the CURRENT state of dev-team, not an older version
• Be honest about gaps — don't dismiss competitors to protect ego
• Be skeptical about install counts — read the actual code/README
• Prefer stealing ideas over adopting whole plugins (less dependency, more control)
• If a plugin does something dev-team does but better, say so explicitly
• Note license compatibility (dev-team is MIT — flag GPL/AGPL conflicts)
• If nothing interesting was found in the time window, say so — don't invent findings