Know what's been built. Find your gap.

An AI tool for Stellar builders. Validate ideas, surface prior art, and match open SCF-funded briefs — whether you're entering a hackathon, applying for a grant, or shipping independently. Works as a SKILL.md or an MCP server.

npx skills add Stellar-Light/stellar-scout

For Codex append -a codex; for OpenClaw, -a openclaw. Installs the SKILL.md into your agent's skills directory. Powered by vercel-labs/skills.

Works with

Claude CodeCodexCursorOpenClawAmpAntigravityCline

Endpoints

4,541

Research chunks

15+

Agent clients

View raw Skill repo MCP repo Browse all Stellar AI tools →

What's inside

Curated Stellar ecosystem data + the Stellar Foundation's official skill catalog — all queryable from inside Claude Code, Codex, Cursor, or any agent that loads SKILL.md.

670+

Stellar projects

Curated projects with category, hackathon history, SCF funding, and GitHub activity. Keyword search across name, description, category.

Sponsor briefs (RFPs)

Prices APIPasskey UI KitDeFi PositionsTrustline Onboarder

Confirmed problem statements that get funded by the Stellar Community Fund when winners are picked. Match an idea to an open brief.

Hackathons

Stellar Hacks: AgentsKALE × ReflectorZK GamingScaffold Stellar

Live DoraHacks feed (SDF + Tellus orgs) + curated Stellar events with prize pools, tracks, winners, and outcome funnels.

Topic clusters

SorobanAnchorsAgentic PaymentsAssetsZK ProofsSEPs

Stellar-native categories aligned to skills.stellar.org's taxonomy. Scout uses these to frame what to build and which SDK to install next.

Companion skills (from SDF)

sorobandappassetsdataagentic-paymentszk-proofsstandards

The Stellar Foundation's 7 official skills — the "how to build" layer. Scout chains into them via /api/skills when the user moves from research to execution.

Two modes

Conversational + Deep Dive

Conversational

Fast, cited answers for any question. Scout hits the right endpoint, surfaces the data, links the source.

Deep Dive

Triggered by "vet this idea" or "should I build X". Runs an 8-step workflow: existing projects → gap classification → competitors → SDK pick → teammates → funding signal → next steps.

The strict part

Gap classification, no speculation

0–2

Full gap

Highest opportunity

Zero prior projects on Stellar. No winning hackathon submissions. No SCF-funded teams in this lane.

3–5

Partial gap

Medium opportunity

1–3 adjacent projects exist but none cover the specific angle. The user's wedge is fresh.

7–10

False gap

Low opportunity

4+ direct competitors or a funded category leader already exists. Recommend a differentiator or reframe.

Evidence floor: if the data doesn't support a claim, Scout says so. It won't invent competitors, prize amounts, or builder profiles. Missing data is reported as "not indexed" — never papered over.

What it can answer

Six core capabilities

[+]
Has anyone built this?
Search for existing Stellar projects, hackathon submissions, and SCF-funded work that overlaps your idea.
[+]
Hackathon results
Winners, placements, prize tracks, post-hack survival rates per event.
[+]
Spot Stellar builders
Search the Stellar Passport directory by skill, location, SCF tier. Scout flags gaps and points to Discord / GitHub when results are thin.
[+]
Idea validation
Full Deep Dive workflow with gap classification and SDK recommendations.
[+]
Dev metrics
Active dev counts, commit volume, country breakdown via the dev-activity skill.
[+]
RFP discovery
Native query of confirmed Stellar RFPs (SCF-funded sponsor briefs) via /api/rfps. Match an idea to an open brief or propose a new one.

Sample prompts

Try these in your agent

Vet a new idea

>"I want to build a privacy-preserving stablecoin on Stellar. Vet this idea."

Find existing projects

>"What Stellar projects have built stablecoin off-ramps? Show me the SCF-funded ones first."

Spot Stellar builders

>"Find Stellar builders who've shipped Soroban contracts (small directory — flag any gaps)."

See all 6 prompts + 5 worked sessions →

Under the hood

Public APIs the skill calls

GET/api/research— vector search over 4.5k chunks

GET/api/hackathons— curated + DoraHacks live

GET/api/projects/search— prior-art + competitor lookup

GET/api/rfps— SCF-funded sponsor briefs

GET/api/skills— SDF skill catalog

GET/api/clusters— topic clusters + crowdedness

GET/api/analyze— cross-event analytics

POST/api/feedback— in-skill feedback loop

…and 6 more

Full API reference (14 endpoints, params, response shapes) →

FAQ

Common questions

Which AI agents does Scout work with?

Any agent that loads SKILL.md files — Claude (claude.ai), Claude Code, Codex, Cursor, OpenClaw, and dozens more (the underlying skills CLI supports 55+ agents). The npx install handles per-agent placement automatically.

Does Scout write code for me?

No. Scout is a research skill — it tells you what's been built, who's building it, what got funded, and which SDK skill to install next. For the actual code work, install the relevant skill from skills.stellar.org (soroban, dapp, assets, etc.) — those are the technical execution layer.

What if Scout returns no results for my query?

It'll tell you so explicitly rather than fabricating answers. Try broader keywords, drop filters (e.g. remove scfAwarded=1), or check that the topic is one we index (curated Stellar projects, hackathons, SCF rounds, RFPs, dev stats). For raw GitHub searches across all of Stellar, use GitHub directly — Scout only covers projects in the stellarlight directory.

How does Scout differ from just asking ChatGPT/Claude about Stellar?

Scout doesn't bring its own brain — it brings structured Stellar data. The skill teaches whatever agent you use (Claude, ChatGPT, etc.) how to query our public APIs to get cited, evidence-backed answers grounded in real numbers: SCF dollars raised, hackathon prize pools, project counts, dev activity. Without Scout the agent guesses; with Scout it cites.

How current is the data?

Varies by source. Project + Hackathons + Builders metadata: refreshed continuously by curators. Live DoraHacks events: cached 1 hour. Ecosystem dev stats (Electric Capital snapshot): daily at 06:00 UTC. SDF skill catalog proxy: 24h. RFPs: live from src/data/ideas.ts. Hit /api/status for exact lastUpdatedAt per source.

Is the API really free? No auth, no rate limits?

Yes — all endpoints are public, read-only, no auth, no rate limits. Edge- cached for 5 minutes (24h for the SDF skill proxy). Hit them from your agent, your Dune query, your dashboard, or anywhere else. If usage ever gets large enough to need rate limiting, we'll publish that ahead of time on the API reference page.

How does Scout relate to skills.stellar.org?

They compose. Scout answers "what should I build, with whom, for what funding?" — strategy. skills.stellar.org (the Stellar Development Foundation's 7 official skills — soroban, dapp, assets, data, agentic-payments, zk-proofs, standards) answers "how do I actually build it?" — execution. Scout cross-links to those skills in its responses; install both when you move from research into building.

What if I find a bug, missing data, or want to suggest a source?

Open an issue in the stellar-scout repo. For missing projects in the directory, use the Submit form at stellarlight.xyz/submit.

What does Scout do when my idea already exists?

It tells you. Scout classifies every idea as a full gap (zero prior projects), partial gap (1–3 adjacent projects), or false gap (4+ direct competitors or a funded category leader), with a crowdedness score (1–10) and a list of the existing players. If the gap is false, it'll suggest where the whitespace actually is rather than just affirming your idea.

What if there's no RFP / sponsor brief matching my idea?

Zero RFPs in a category doesn't mean "no opportunity" — it means no sponsor brief in the current SCF round covers that lane yet. Anyone can propose an RFP at stellarlight.xyz/ideas via the "Suggest a Need" button. Community submissions go through curators and graduate to confirmed RFPs in upcoming rounds. Scout will tell you this directly when its RFP search comes up empty.

Ready to validate your hackathon idea?

Install Scout, ask "should I build X on Stellar?" — get evidence.