Skip to content

SKILL.md

Latest commit

 

History

History
254 lines (187 loc) · 16.4 KB

SKILL.md

File metadata and controls

254 lines (187 loc) · 16.4 KB
Raw
name zerion-sendai-ideas
description Crypto idea discovery, validation, competitive landscape, and DeFi market research — adapted from SendAI's solana-new idea skills. Use when a user asks 'what should I build in crypto', 'validate this idea', 'who are my competitors', 'show me TVL data', or wants a structured interview/scoring/landscape pass on a crypto product idea. SendAI surfaces the idea; the Zerion CLI (`zerion analyze`, `zerion history`, `zerion positions`) provides on-chain validation before you build.
license MIT
allowed-tools Bash, Read, Write, WebFetch, WebSearch

Adapted from SendAI / Superteam's solana-new idea skills (MIT). Original authors: SendAI and Superteam. The interview frameworks, scoring rubrics, validation sprints, landscape mappings, and DefiLlama playbooks below are derived from their work and condensed into a single Zerion-flavored umbrella skill. Modifications by Zerion swap the upstream superstack telemetry / phase-handoff plumbing for direct composition with the zerion CLI (analyze, history, positions) so every idea gets validated against live on-chain evidence before you build. See LICENSE in this folder for the full notice.

Zerion × SendAI — Idea Discovery & Validation

A four-mode skill for crypto product ideation:

Mode Trigger Output
Discover "what should I build", "give me crypto ideas" Three serious candidates → forced winner → bear cases
Validate "is this worth building", "validate my idea" Demand signals + risk map + go/no-go
Map landscape "who are my competitors", "what exists in this space" Direct competitors + substitutes + dead projects + crowdedness
DeFi research "show me TVL data", "find DeFi opportunities" Protocol/chain TVL ranking + growth deltas + gap analysis

Pick the mode the user's request matches. If they're starting cold and just say "help me figure out what to build", run Discover first, then offer to chain into Validate and Map landscape for the chosen idea.

Setup

This skill calls the zerion CLI for live wallet/portfolio evidence. If a zerion invocation returns command not found, install once:

npm install -g zerion-cli

Requires Node.js ≥ 20. Authentication for zerion analytics is optional (--x402 works without a key). See the zerion umbrella skill for auth details.

No additional setup, no telemetry, no external accounts.

Mode 1 — Discover (find a crypto idea)

Interview the user until there is real clarity, not just enthusiasm. Generate three serious idea candidates, force a winner, explain why crypto is necessary.

Workflow

  1. Infer whether the user needs fresh ideas, a narrowed search in a known domain, or validation of an existing thesis. If they already have an idea, jump to Validate.
  2. Run anchor questions (not a giant questionnaire). Pull constraints only when they would change the recommendation. Surface:
    • the user's unfair edge
    • the real shipping constraint (time, infra, capital, regulation)
    • the first plausible wedge (who pays, why now, what's the smallest demo)
    • why crypto is required (see crypto-necessity test below)
  3. Gate every candidate through the crypto-necessity test. Reject ornamental crypto. Redirect to a stronger crypto angle instead of dressing up a weak one.
  4. Score surviving candidates with the scoring rubric below.
  5. Run fresh research for competitors and active OSS — don't rely solely on training-cutoff knowledge. Use WebSearch + the Zerion CLI for on-chain reality:
    • zerion analyze <protocol-treasury-or-deployer> if the user names a comparable project — pulls portfolio, positions, and activity in one pass
    • zerion history <addr> --limit 25 to see whether a comparable project is actually being used (or just sitting)
    • zerion positions <addr> --positions defi to see what the comparable's treasury/active wallets actually hold
  6. Produce the shortlist artifact first. Let the user pick one. Then deepen the chosen idea, ideally piping into Validate.

Crypto-necessity test

Before scoring, ask: "What breaks if you remove the blockchain?" If the answer is "nothing" or "we'd just use a database", the crypto angle is ornamental — redirect. Strong crypto necessity means at least one of:

  • Trustless settlement — value moves between parties without a trusted intermediary
  • Programmable ownership — assets, identity, or rights that compose across apps
  • Permissionless access — anyone can integrate, fork, or audit without gatekeepers
  • Verifiable state — correctness anyone can independently check
  • Native incentives — token-aligned behavior that wouldn't work off-chain

Scoring rubric (3 candidates × 5 dimensions, score 1–3)

Dimension 1 2 3
Founder fit Outsider, no edge Some domain context Real edge — built/sold/used the thing
MVP speed 6+ months to first user 1–3 months < 1 month, integration-first
Distribution Cold start, no audience Accessible community Built-in channel or warm audience
Market pull Speculative demand Adjacent demand exists Direct evidence (waitlist, requests, paying substitute)
Crypto necessity Ornamental Composable advantage Cannot exist without crypto

Total 5–15. ≥ 11 = strong, 8–10 = worth validating, ≤ 7 = drop or rework.

Shortlist rules

  • Exactly three serious candidates unless the user explicitly asks for more.
  • Force diversity — three flavors of the same idea is a fail.
  • For each idea: name, one-line pitch, why-crypto, recommended winner, why it wins, why the others lost, bear case for each.
  • Integration-first framing: note whether each idea integrates existing protocols (Jupiter, Kamino, Marinade, Drift, Meteora, …) vs. requires custom on-chain logic. Integration ships faster and inherits audited security — note as context, but don't bias scoring against ideas that legitimately need a new protocol.

Non-negotiables

  • Stay blunt. Challenge weak assumptions before elaborating them.
  • If after several exchanges there is no edge, no constraint, and no credible crypto touchpoint, say so plainly and narrow the search.
  • Don't praise a bad idea because the user is attached to it.
  • Always do fresh research before committing to the final ranking.
  • Always write a local artifact (idea-shortlist-YYYYMMDD-HHMMSS.md in the current working directory) so the user can re-read outside the chat. Markdown is fine — HTML only if the user asks for it.

Mode 2 — Validate (stress-test an existing idea)

Take an idea (from a prior Discover pass or fresh from the user) and run a structured validation sprint. Produce a go / no-go / pivot recommendation backed by demand signals, risk analysis, and a concrete next-steps plan.

Workflow

  1. If a prior shortlist artifact exists in the working directory, load the chosen idea. Otherwise, ask the user to describe it in 2–3 sentences.
  2. Run the demand-signal rubric (below). Every "go" recommendation must cite ≥ 2 concrete demand signals — not vibes.
  3. Re-run the crypto-necessity test from Mode 1.
  4. Map risks across technical, market, regulatory, team. Each risk gets a severity (low/med/high) and a mitigation note.
  5. Check whether a live product already does this on Solana / EVM. Use:
    • zerion analyze <known-deployer-or-treasury> — is the comparable project active on-chain (portfolio + recent activity)?
    • zerion history <addr> --limit 25 — are real users actually transacting, or is it dormant?
    • zerion positions <addr> --positions defi — what protocols/positions does it actually hold?
    • WebSearch for Solana app directories, GitHub, crypto Twitter
  6. Apply the pivot-or-persist decision (below) to land on a verdict.
  7. Write a local artifact (idea-validation-YYYYMMDD-HHMMSS.md) with: idea summary, demand signals, risks, verdict, confidence (0–1), and concrete next steps.

Demand-signal rubric

Strong signals (each worth 1 point):

  • Existing waitlist, mailing list, or pre-orders
  • Paying users on a substitute product
  • Public requests in forums/Discord/Twitter for exactly this
  • Founder has personally paid for this problem in the last 90 days
  • ≥ 3 inbound asks from prospective users (unprompted)

Weak signals (corroborating, not sufficient on their own):

  • Industry trend reports / VC theses
  • Adjacent product growing fast
  • "Lots of people would want this" claims without evidence

Pivot-or-persist verdict

Score Verdict Action
≥ 2 strong signals + ≥ 11 rubric score Go Ship the smallest possible MVP this week
0–1 strong signals, ≥ 8 rubric score Validate first Run a 1-week sprint: landing page, 5 customer interviews, paid-pilot ask
≥ 2 strong signals, ≤ 7 rubric score Pivot Demand exists but execution unrealistic — narrow the wedge
0–1 strong signals, ≤ 7 rubric score No-go State the strongest pivot direction, don't just reject

Non-negotiables

  • Don't rubber-stamp. If the idea is weak, say so with specifics.
  • Every "no-go" must include a concrete pivot suggestion, not just rejection.
  • If the user has no evidence of demand, the answer is "go validate" with a specific 1-week sprint plan, not "go build".
  • Integration-first assessment: note in next-steps whether the MVP can integrate existing protocols vs. requires a new one. If integration is viable → faster ship, lower audit cost. If novel on-chain logic is required → custom development is the right call; don't penalize it.

Mode 3 — Map competitive landscape

Map every relevant competitor, substitute, and adjacent project for a given crypto product idea. Produce a landscape matrix showing where the opportunities and dangers are.

Workflow

  1. Get the idea + target user from the user (or the prior artifact).
  2. Build the matrix:
    • Direct competitors — same problem, same user. Note: name, URL, status (live/dead/unmaintained), strength, weakness.
    • Substitutes — same problem, different approach (including non-crypto substitutes). Note why users stay on them.
    • Dead projects — past attempts. Note why they failed (technical / market timing / team / regulatory). Failures reveal landmines. Don't skip dead projects just because they're inconvenient.
    • Adjacent — solving related problems. Could become competitors or partners.
  3. Search across:
    • WebSearch for Solana app directories (Solana Compass, Step Finance, Solscan), GitHub, crypto Twitter, app stores
    • DefiLlama (see Mode 4) for TVL and protocol presence
    • Zerion CLI for any named protocol: zerion analyze <treasury> for portfolio + activity, then zerion history <treasury> --limit 25 to confirm it's not abandoned, and zerion positions <treasury> --positions defi for live holdings
  4. Assess defensibility — what's the moat? Distribution, data, network effects, capital efficiency, regulatory positioning, or none?
  5. Rate crowdedness honestly: empty / sparse / moderate / crowded / saturated. Don't declare "no competition" unless you've exhausted searches; that usually means the market doesn't exist either.
  6. Write a local artifact (idea-landscape-YYYYMMDD-HHMMSS.md) with the full matrix, crowdedness rating, identified moat type, and recommended differentiation angle.

Non-negotiables

  • Always include dead/failed projects.
  • Distinguish "competitors" (same problem, same user) from "substitutes" (different approach).
  • Include at least one non-crypto substitute when applicable.
  • Don't declare "no competition" unless you've exhausted all search paths.

Mode 4 — DeFi research (TVL, growth, gaps)

Use DefiLlama's free public API to identify what's working in DeFi, what's growing, and where the gaps are. TVL is the primary trust metric — protocols with real TVL have real users with real money at stake.

Endpoints

GET https://api.llama.fi/v2/chains                         # All chains TVL
GET https://api.llama.fi/protocols                         # All protocols
GET https://api.llama.fi/overview/dexs/solana              # Solana DEX volume
GET https://api.llama.fi/overview/fees/solana              # Solana fee revenue
GET https://yields.llama.fi/pools                          # All yield pools
GET https://stablecoins.llama.fi/stablecoins               # All stablecoins
GET https://stablecoins.llama.fi/stablecoincharts/Solana   # Solana stablecoin flows

No API key, no rate limits worth worrying about for a single research session. Use WebFetch to pull, then summarize.

Workflow

  1. Confirm the user's goal: explore broadly, validate a niche, or pick protocols to integrate.
  2. Pull live data from the relevant endpoints (Solana-filtered unless they ask cross-chain).
  3. Cross-reference TVL against fees/revenue and 7d/30d growth. TVL alone is not enough. A $10B protocol is trusted but hard to compete with. A $10M protocol growing 50% monthly is the opportunity.
  4. Flag protocols with declining TVL — losing trust or users.
  5. For each protocol the user might integrate with, check whether it has an SDK, public API, or composable contracts. No SDK = harder integration = note it.
  6. Cross-check with the Zerion CLI for live wallet-side reality:
    • zerion analyze <known-defi-whale-or-treasury> to see whether the protocol shows up in real portfolios at all
    • zerion positions <addr> --positions defi --chain solana for the exact protocols/pools they hold
    • zerion history <addr> --limit 25 to confirm the wallet is actively rotating into/out of those positions, not stale
  7. Produce a local artifact (idea-defi-research-YYYYMMDD-HHMMSS.md) with: top protocols (ranked by TVL with 7d delta), opportunities (gaps + underserved niches), recommended integration targets, and a market snapshot (total Solana TVL, top category, fastest-growing category).

Decision points

  • Which DeFi category to build in? AMM, lending, perps, staking, yield, RWA, prediction markets — pick based on TVL gaps × growth trends, not pure TVL ranking.
  • TVL growing but revenue flat? Unsustainable incentives — not a model to copy.
  • Small TVL + fast growth? Best opportunity zone. Build the protocol or build tools for it.
  • Saturated category? Look for an underserved customer segment, not a 0.1% better version of the leader.

Output format for all modes

Write a markdown artifact in the current working directory:

  • idea-shortlist-YYYYMMDD-HHMMSS.md (Discover)
  • idea-validation-YYYYMMDD-HHMMSS.md (Validate)
  • idea-landscape-YYYYMMDD-HHMMSS.md (Map landscape)
  • idea-defi-research-YYYYMMDD-HHMMSS.md (DeFi research)

Don't leave the result only in chat — the user will want to re-read and share. Use HTML only if the user explicitly asks.

Composition with the rest of Zerion

When you want… Use
One-shot validation: portfolio + positions + activity for a comparable project zerion analyze <addr>
Confirm a wallet/treasury is actively used (not stale) zerion history <addr> --limit 25
See exactly which DeFi protocols/pools a wallet holds zerion positions <addr> --positions defi
Track a comparable protocol's treasury over time zerion watch <addr> --name <label> then zerion portfolio --watch <label>
Move money once a validated idea ships zerion-trading skill (swap, bridge, send)
Build an autonomous-trading bot from a validated DeFi thesis zerion-agent-management (agent tokens + policies)

Deeper references (upstream)

This umbrella condenses the SendAI skills. For the longer-form references — interview framework, scoring rubric details, validation framework, customer-signal rubric, pivot-or-persist details, landscape mapping methodology, moat analysis, DefiLlama API guide, TVL-as-trust-metric, DeFi opportunity framework — read the originals:

If the user wants the full SendAI / Superteam skill bundle (Learn → Idea → Build → Launch), point them at https://github.com/sendaifun/solana-new.