Ecosystem Scout — Discover and Catalog AI Tools

Discover new AI tools, evaluate them, update the ecosystem catalog, and generate recommendations for active ventures.

Input

Focus area (optional): $ARGUMENTS

• If provided: focused sweep on that area (e.g., "mcp-servers", "marketplace-tools", "content-generation")
• If empty: broad sweep across all discovery sources

Step 1: Read Current State

1. Read knowledge/ecosystem/_index.md for the current catalog
2. Read knowledge/ecosystem/_categories.md for taxonomy and scoring guide
3. Read knowledge/ecosystem/_recommendations.md for previous recommendations
4. Read portfolio/registry.md for active ventures and their stages
5. Read knowledge/lessons.md for relevant learnings

Note: Remember the current catalog entries so you can detect what's new vs. already cataloged.

Step 2: Discovery Sweep

Spawn the scout agent to search for tools.

If focus area provided ($ARGUMENTS):

• Search specifically for tools related to $ARGUMENTS
• Use targeted queries: "{focus-area} AI tools", "{focus-area} open source", "{focus-area} MCP server"

If broad sweep (no arguments):

• Search GitHub trending for AI/ML repos (last week)
• Search for new MCP servers and integrations
• Search for new AI agent frameworks and tools
• Search for tools relevant to active venture types
• Check awesome-lists for recently added entries (awesome-mcp, awesome-ai-agents, etc.)

For each discovery, collect:

• Name, URL, description
• GitHub stars (if applicable), last commit date
• License
• Category and tags (per _categories.md)

Target: 5-15 discoveries per sweep (quality over quantity).

Step 3: Evaluate Discoveries

For each discovered tool that isn't already in the catalog (or has changed significantly):

1. Fetch the tool's homepage/README to understand capabilities
2. Score across 5 dimensions (see _categories.md for scoring guide):
   • Adoption (0-20)
   • Recency (0-20)
   • Documentation (0-20)
   • Relevance (0-20)
   • Community (0-20)
3. Classify by category and tags
4. Assess venture relevance — which venture types and stages would benefit?
5. Write entry file to knowledge/ecosystem/entries/{slug}.md

Skip tools that score below 30 total — not worth cataloging.

Step 4: Re-verify Existing Entries

Check entries in the catalog that were last verified more than 30 days ago:

1. Read the entry file
2. Quick web check — is the project still active? Any major changes?
3. Update score if the situation has changed
4. Mark status:
   • active — still maintained and relevant
   • stale — no updates in 6+ months, reduced score
   • deprecated — project archived or abandoned, set score to 0
   • archived — we've decided not to track this anymore

Step 5: Rebuild Index

Rewrite knowledge/ecosystem/_index.md with all current entries:

• Include all entries with status active or stale
• Sort by score (highest first)
• Exclude deprecated and archived entries from the main table (list them in a separate section)

Step 6: Generate Scout Report

Output a report summarizing:

New Discoveries

Updated Entries

Deprecated/Stale

Top Recommendations for Active Ventures

For each active venture, list the top 3 most relevant tools with integration suggestions.

Step 7: Write Recommendations

Rewrite knowledge/ecosystem/_recommendations.md with current recommendations:

For each active venture in the portfolio:

1. Filter catalog for entries matching the venture's type and stage
2. Only include entries with Score >= 60
3. Classify each recommendation:
   • Agent Enhancement — improves an existing venture agent
   • Skill Component — can be incorporated into a workflow
   • Future-Stage Addition — valuable at a later stage
   • Reference Only — worth knowing about, no immediate action
4. Include integration notes for each

Update the "Last Updated" and "Active Ventures Considered" fields.