Add Messari x402 research skills + prediction market scout

[tonyagents]

Skill name

messari-x402, messari-token-research, messari-alpha-scout, messari-funding-intel, messari-deep-research

Description

Five Messari x402 research skills — pay-per-request access to Messari's full crypto intelligence API via USDC on Base. No API key required; payments handled automatically via MoonPay x402.

• messari-x402 — Core reference skill: endpoint catalog, cost table, auth pattern
• messari-token-research — Full token research workflow (fundamentals + price history + sentiment + AI synthesis)
• messari-alpha-scout — Mindshare gainers, trending topics, and emerging narrative detection
• messari-funding-intel — VC funding rounds, M&A activity, venture landscape brief
• messari-deep-research — Async long-form research reports (polls until complete)

Scout skill removed from this PR — lives in its own PR #8 as moonpay-scout.

Primary chain

Base (USDC payments via x402)

Primary token

USDC (Base)

Checklist

• Frontmatter (name, description, tags) present on all 5 skills
• Skills registered in .claude-plugin/marketplace.json under messari-skills plugin
• Descriptions are specific about when Claude should trigger each skill
• Naming convention: all use messari-{name}/ format
• No fabricated commands — all mp x402 request calls use verified Messari endpoints
• Cross-references to related skills included
• Portable date command (works on macOS and Linux)
• Cron scheduling uses consistent crontab (not mixed launchd/crontab)

MoonPay Integration

All skills use mp x402 request to make pay-per-request API calls. USDC on Base is auto-deducted from the main wallet per call. No separate auth or API key needed — MoonPay handles payment.

Example Usage

Run a full token research workflow on Bitcoin (~$1.00-1.50 USDC total):

mp x402 request --method GET --url https://api.messari.io/v1/assets/bitcoin/metrics --wallet main --chain base

Start an async deep research report:

mp x402 request --method POST --url https://api.messari.io/ai/v0/research/create --body '{"query":"What is the investment thesis for Ethereum restaking?"}' --wallet main --chain base

[kevarifin14]

PR Review: Add Messari x402 research skills + prediction market scout

I reviewed all 6 new skills and the marketplace.json changes against the repo's conventions in CLAUDE.md, the skill template, and the PR template. Here is a detailed breakdown.

---

1. PR Template Compliance

The PR template at .github/PULL_REQUEST_TEMPLATE.md expects specific fields: Skill name, Description, Primary chain, Primary token, Checklist (frontmatter, marketplace.json, description specificity), MoonPay Integration, and Example Usage. This PR does not follow the PR template at all -- it uses a freeform Summary/Test plan format instead. The checklist items from the template are not checked off.

Action required: Restructure the PR description to follow the template, or at minimum address the checklist items explicitly.

---

2. Naming Convention Issues

Per CLAUDE.md: partner skills use skills/{partner}-{name}/ format.

• skills/messari-x402/ -- Correct (partner=messari, name=x402).
• skills/messari-token-research/ -- Correct.
• skills/messari-alpha-scout/ -- Correct.
• skills/messari-funding-intel/ -- Correct.
• skills/messari-deep-research/ -- Correct.
• skills/scout/ -- VIOLATION. This is not a MoonPay skill (moonpay-scout) nor does it follow partner naming ({partner}-{name}). What partner does this belong to? If it is a MoonPay-authored utility, it should be moonpay-scout or moonpay-prediction-arb. If it is from a specific partner, it needs that prefix.

Action required: Rename skills/scout/ to follow the naming convention.

---

3. Frontmatter Review

messari-x402:

• name: present, correct.
• description: Good -- specific about what it does and how (x402, USDC on Base, no API key). Adequate trigger guidance.
• tags: present.

messari-token-research:

• name: present, correct.
• description: Good -- mentions the workflow steps and cost. Could be slightly more specific about when to trigger (e.g., "Use when the user asks to research a specific token").
• tags: present.

messari-alpha-scout:

• name: present, correct.
• description: Good, mentions what it scans and cost.
• tags: present.

messari-funding-intel:

• name: present, correct.
• description: Good, specific about the workflow.
• tags: present.

messari-deep-research:

• name: present, correct.
• description: Good, mentions async nature and cost range.
• tags: present.

scout:

• name: present but see naming issue above.
• description: MAJOR ISSUE. The description is just "Prediction Market Arbitrage & Alpha Scout" -- this is a title, not a description. Per CLAUDE.md: "The description field in YAML frontmatter should be specific about when Claude should trigger the skill -- this is how agents decide which skill to load." An agent reading this description has no idea when to load this skill. It needs something like: "Use when the user wants to find arbitrage opportunities between Polymarket and Kalshi prediction markets, or asks to scan for mispricings across prediction market platforms."
• tags: present.

Action required: Fix the scout description to be specific about when Claude should trigger it.

---

4. marketplace.json Review

• Messari skills are correctly grouped under a messari-skills plugin entry with all 5 skills listed. Good.
• Scout is in its own plugin entry. Fine structurally.
• However: The messari-skills plugin source is "./" which matches the existing moonpay-skills entry. Verify this is intentional and that the plugin loader handles multiple plugins with the same source directory.

---

5. Skill Content Quality -- Messari Skills

Overall: The five Messari skills are well-structured, detailed, and genuinely useful. They follow the template pattern (Goal, Prerequisites/Preflight, Steps, Workflow script, Output format, Notes, Related skills). Specific feedback:

messari-x402 (core reference):

• Excellent endpoint reference table with costs. This is a high-quality reference skill.
• Has prerequisites, examples, and a discovery endpoint. Good.
• Minor: The template expects a "Prerequisites" section with mp login / mp verify -- this skill has it informally but does not mention mp login or mp verify explicitly. Should it? If auth is handled by x402 flow, state that explicitly.

messari-token-research:

• Clear 5-step workflow with cost breakdown per step. Good.
• Step 2 uses date -v-30d which is macOS-specific (-v flag). On Linux this would be date -d '30 days ago'. Since this is an agent-executed skill, the agent may run on either platform. Consider noting this or using a portable alternative.
• The "Trigger phrases" section is helpful for agent routing -- good pattern, though this info should also be captured in the description frontmatter.

messari-alpha-scout:

• Well-structured 4-step scan + AI synthesis.
• Includes a cron scheduling section which is a nice touch.
• Minor: Says "macOS launchd" but then shows crontab instead of a launchd plist. Pick one.

messari-funding-intel:

• Solid workflow. Step 4 is marked optional which is good.
• Good cross-referencing to other Messari skills.

messari-deep-research:

• Async polling pattern is well documented with status values, timeout guidance, and cancel endpoint.
• Good example research queries section.
• The polling script uses python3 -c for JSON parsing. Consider noting jq as an alternative or requirement.

---

6. Skill Content Quality -- Scout

This is the most problematic skill in the PR. Several issues:

a) No mp CLI commands at all. The skill references Polymarket and Kalshi but never shows how to actually call their APIs. There are no mp commands, no curl commands, no API endpoints -- just prose descriptions of what to do. An agent loading this skill would not know how to execute any of the steps. Compare this to the Messari skills which give exact mp x402 request commands for every step.

Specifically:

• Step 1 says "search both Polymarket and Kalshi" but doesn't say how. What API? What mp command?
• Step 2 says "extract" prices but from what response?
• Step 5 says "place the position using the correct tokenId" but gives no command.

This reads more like a strategy document than an executable skill. Skills are "instructional guides" but they need to be actionable -- an agent needs concrete commands to run.

b) No Prerequisites section. The template expects prerequisites (CLI installed, authenticated, funded wallet). The scout skill jumps straight into steps with no setup guidance.

c) No cost information. Unlike the Messari skills which carefully document costs per step, the scout skill has no cost information for trades or API calls.

d) Hardcoded assumptions. The skill assumes $10 position size and main wallet without explaining these are configurable or where they come from.

e) The "Iran lesson learned" line (last line of the file). This is oddly specific anecdotal knowledge baked into a general-purpose skill. It reads like a personal note rather than a generalizable instruction. Either remove it or generalize it (e.g., "When volumes differ significantly between platforms, the higher-volume platform is likely more informed on that specific event").

f) Uses emojis extensively in output templates. Not a hard blocker but inconsistent with the more professional tone of the Messari skills.

Action required: The scout skill needs significant rework:

1. Add concrete API calls or mp commands for scanning Polymarket and Kalshi
2. Add a Prerequisites section
3. Add cost/fee information
4. Generalize the "Iran lesson learned"
5. Fix the frontmatter description (see issue #3 above)
6. Fix the naming (see issue #2 above)

---

7. Potential Hallucinated APIs

The Messari API endpoints look plausible and are internally consistent across skills (same base URL, same auth pattern, costs referenced in the x402 skill match costs in workflow skills). The /.well-known/x402 discovery endpoint is a nice verification mechanism. I cannot confirm these endpoints are real without testing, but they are at least self-consistent.

The scout skill references no specific APIs at all, so there is nothing to verify there -- which is itself a problem.

---

8. Cross-References

The Messari skills cross-reference each other well and also reference relevant MoonPay skills (moonpay-swap-tokens, moonpay-check-wallet, moonpay-budget-agent, moonpay-auth). Good.

The scout skill has zero cross-references to other skills. It should at minimum reference moonpay-fund-polymarket and moonpay-prediction-market which already exist in the repo.

---

Summary

Skill	Verdict
messari-x402	Approve -- high quality reference skill
messari-token-research	Approve with minor fixes -- macOS-specific date command
messari-alpha-scout	Approve with minor fixes -- crontab/launchd inconsistency
messari-funding-intel	Approve -- solid
messari-deep-research	Approve -- solid
scout	Request changes -- needs naming fix, description fix, actual API commands, prerequisites, cross-references

Recommendation: Split the scout skill into a separate PR and land the Messari skills (with minor fixes) now. The scout skill needs more work to meet the quality bar set by the Messari skills in this same PR.

[kevarifin14]

Grade: B+ (Messari skills) / D (Scout skill)

Messari skills (5 of them): B+ — These are the highest quality submissions in the batch. Clear mp x402 request commands, good cost breakdowns, proper cross-references. Close to A but not quite — minor issues (macOS-specific date -v, crontab/launchd inconsistency) and the scout skill dragging down the PR.

Scout skill: D — No executable commands, reads as a strategy doc not a skill. Vague description, missing prerequisites, no cross-references to existing prediction market skills.

Recommendation: Split out the scout skill, fix the minor issues on the Messari skills, and this is mergeable. The Messari skills on their own would be an A-.

[kevarifin14]

Re-review after fixes (4 new commits)

All 3 flagged issues addressed:

• ✅ Scout skill split out (separate PR #8)
• ✅ macOS date -v replaced with portable fallback
• ✅ crontab/launchd inconsistency fixed

Verification results:

• mp x402 request — confirmed real (mp x402 --help)
• clawhub — confirmed real (npm, v0.8.0)
• All Messari API endpoints use mp x402 request which handles auth — valid pattern

Updated grade: A-

One remaining blocker to get to A+:

1. marketplace.json placement — The 5 Messari skills are added inside the moonpay-skills plugin array. Partner skills should have their own plugin block:

{
  "name": "messari-skills",
  "description": "Messari crypto research and analytics via x402 micropayments",
  "source": "./",
  "strict": false,
  "skills": [
    "./skills/messari-x402",
    "./skills/messari-token-research",
    "./skills/messari-alpha-scout",
    "./skills/messari-deep-research",
    "./skills/messari-funding-intel"
  ]
}

Fix that and this is ready to merge.

[kevarifin14]

Grade: A+

All issues resolved:

• ✅ Scout skill split out
• ✅ Messari skills in own messari-skills plugin block
• ✅ Em-dash diff clean — no unrelated changes
• ✅ Portable date command (GNU + macOS fallback)
• ✅ Consistent cron scheduling

Verified: mp x402 request (real), clawhub (npm v0.8.0). All 5 Messari skills are high quality with clear workflows, cost breakdowns, and cross-references.

Ready to merge.