| name | stellar-competitive-landscape |
| description | Map the competitive landscape for a Stellar project idea. Use when a user says "who are my competitors on Stellar", "competitive analysis Stellar", "what already exists in this space on Stellar", "show me similar Stellar projects", "is anyone else building this on Stellar", or "map the Stellar landscape for X". Queries the 728-project LumenLoop ecosystem database and Electric Capital developer activity to rank competitors by SCF funding history and repo activity. |
What this skill does
1. Get the project context
If the user hasn't described their idea, ask: "What are you building, in one sentence?"
Then ask: "Any specific category to compare against? — Financial Protocols, Applications, Developer Tooling, Infrastructure & Services, Payments, Education & Community, or other."
2. Search the ecosystem DB
Read ~/.claude/skills/data/lumenloop/projects.json. If missing, fetch from https://raw.githubusercontent.com/lumenloop/stellar-ecosystem-db/main/ (the YAML files in projects/ — convert on the fly if needed).
Filter by:
attributes.category matching the user's stated category (case-insensitive partial match)- Description keyword overlap with the user's idea — extract 3-5 keywords from the idea, score each project by how many keywords appear in its description
Return top 10-15 matches ranked by relevance score.
2.5 Mine Electric Capital for what LumenLoop doesn't show
~/.claude/skills/data/electric-capital/stellar-repos.json has ~9000 repos tagged as Stellar — a much wider net than LumenLoop's curated 728. Use it to surface what LumenLoop misses:
- Emerging players: GitHub orgs with 2+ Stellar repos that aren't in LumenLoop yet (proto-projects, soon-to-launch)
- Dormant competitors: repos whose names match the user's keywords but show no recent activity (these were attempts that didn't ship — useful signal about why the category might be hard)
- Cluster detection: if 5+ different orgs are working on the user's exact keyword (e.g., "perps", "vault", "wallet"), the space has more friction than LumenLoop suggests
Filter out noise:
- Repos with
whitebelt, orangebelt, yellowbelt, greenbelt, redbelt in the name (Stellar's tiered training program)
- Forks of
stellar/* official repos with no meaningful divergence
- Student/tutorial repos (
-tutorial, -exercise, -demo suffixes)
The cleaned signal vs LumenLoop catalog gives you the real picture: who's documented, who's stealth-building, who tried and abandoned.
3. Enrich with developer activity
For each top match, look up its GitHub org/repo in ~/.claude/skills/data/electric-capital/stellar-repos.jsonl. Note: last-commit recency, total contributors if available. Flag dormant repos (no commits in 6+ months) as "low activity."
4. Build the competitive map
Present as a markdown table, sorted by SCF funding desc, then dev activity desc:
| Project | Category | SCF rounds + total | GitHub activity | One-line description | Overlap with user's idea |
|---|
Use scf.awarded_total from the project entry. If null, mark as "—" (community-listed but not SCF-funded).
5. Differentiation guidance
After the table, write 2-3 sentences answering:
- Strongest competitor: which project is closest to the user's idea, what they do, and where their visible gap is (missing feature, dormant repo, narrow scope)
- Saturated vs open space: does the category have 1-2 dominant players (saturated) or many small players (fragmented, possibly open). Saturation is judged on SCF-funded count, not community-listed count.
- Recommended differentiation: one concrete angle the user could lean into
Close with: "Want me to route you to validate-stellar-idea to stress-test against the top competitor, or back to find-stellar-idea to pivot the angle?"
Constraints
- Only count SCF-funded projects toward "saturation" — unfunded community-listed projects are weaker competitive signal
- If user's category isn't recognized, ask for clarification rather than guessing
- If
projects.json returns 0 matches, the category may be genuinely open — surface that as a finding, not a failure
- Do not invent project names. If the DB doesn't have something, say so honestly
