Messari Deep Research Workflow

Goal

Generate a comprehensive, long-form research report on any crypto topic, asset, sector, or question. Uses Messari's async AI deep-research engine which crawls their knowledge graph, on-chain data, and news to produce institutional-quality reports.

Cost per run: ~$0.50–$2.00 USDC on Base (async, takes 5–15 minutes)

Trigger phrases

• "Write a deep research report on [topic]"
• "Give me a full analysis of [asset/protocol]"
• "Research the [DeFi/L2/RWA/etc.] sector"
• "Deep dive on [topic]"
• "Generate a research report"

---

Step 0 — Preflight: check balance

mp token balance list --wallet main --chain base --json

Ensure at least $3.00 USDC on Base.

---

Step 1 — Start deep research job

mp x402 request \
  --method POST \
  --url "https://api.messari.io/ai/v1/deep-research" \
  --body '{
    "query": "<your research question or topic>"
  }' \
  --wallet main \
  --chain base

Example queries:

• "What is the investment thesis for Ethereum restaking protocols in 2025?"
• "Analyze the competitive landscape of L2 rollups"
• "What are the key risks in the RWA tokenization sector?"
• "Deep dive on Solana's DeFi ecosystem"

Response: Returns a job id (e.g. dr_abc123) and status pending.

{
  "id": "dr_abc123",
  "status": "pending",
  "created_at": "2025-..."
}

---

Step 2 — Poll for completion (free, repeat until done)

mp x402 request \
  --method GET \
  --url "https://api.messari.io/ai/v1/deep-research/<JOB_ID>" \
  --wallet main \
  --chain base

Status values:

• pending — job queued, try again in 60s
• processing — actively generating, try again in 60s
• completed — report ready, extract result
• failed — job failed, check error field

Polling is free (no payment required). Poll every 30–60 seconds.

---

Step 3 — Retrieve completed report

Once status is completed, the response contains the full report:

{
  "id": "dr_abc123",
  "status": "completed",
  "result": {
    "title": "...",
    "summary": "...",
    "sections": [...],
    "sources": [...]
  }
}

---

Full workflow script (with auto-polling)

#!/bin/bash
# messari-deep-research.sh "<research query>"
# Usage: ./messari-deep-research.sh "Analyze the Solana DeFi ecosystem"

QUERY="${1:-What are the top crypto narratives for 2025?}"
WALLET="main"
CHAIN="base"
BASE="https://api.messari.io"
OUT="$HOME/.config/moonpay/research/deep-$(date -u +%Y%m%d-%H%M%S)"
mkdir -p "$(dirname "$OUT")"

echo "Starting deep research: \"$QUERY\""
echo ""

# Step 1: Start job
RESPONSE=$(mp x402 request --method POST \
  --url "${BASE}/ai/v1/deep-research" \
  --body "{\"query\": \"${QUERY}\"}" \
  --wallet "$WALLET" --chain "$CHAIN")
echo "$RESPONSE" > "${OUT}-job.json"

JOB_ID=$(echo "$RESPONSE" | python3 -c "import sys,json; print(json.load(sys.stdin)['id'])" 2>/dev/null)

if [ -z "$JOB_ID" ]; then
  echo "ERROR: Could not extract job ID. Response:"
  echo "$RESPONSE"
  exit 1
fi

echo "Job started: $JOB_ID"
echo "Polling for completion (this takes 5–15 minutes)..."

# Step 2: Poll loop
while true; do
  sleep 30
  STATUS_RESPONSE=$(mp x402 request --method GET \
    --url "${BASE}/ai/v1/deep-research/${JOB_ID}" \
    --wallet "$WALLET" --chain "$CHAIN")

  STATUS=$(echo "$STATUS_RESPONSE" | python3 -c "import sys,json; print(json.load(sys.stdin).get('status','unknown'))" 2>/dev/null)
  echo "  Status: $STATUS ($(date +%H:%M:%S))"

  if [ "$STATUS" = "completed" ]; then
    echo "$STATUS_RESPONSE" > "${OUT}-report.json"
    echo ""
    echo "Report complete! Saved to ${OUT}-report.json"
    break
  elif [ "$STATUS" = "failed" ]; then
    echo "ERROR: Research job failed."
    echo "$STATUS_RESPONSE"
    exit 1
  fi
done

---

Cancel a running job

mp x402 request \
  --method POST \
  --url "https://api.messari.io/ai/v1/deep-research/<JOB_ID>/cancel" \
  --wallet main \
  --chain base

---

List past research jobs

mp x402 request \
  --method GET \
  --url "https://api.messari.io/ai/v1/deep-research" \
  --wallet main \
  --chain base

---

Output format

Present the completed report as:

## Deep Research: [Report Title]
**Job ID:** dr_abc123
**Generated:** [timestamp]
**Cost:** ~$X.XX USDC

### Executive Summary
[2-3 paragraph summary]

### [Section 1 Title]
[content]

### [Section 2 Title]
[content]

...

### Sources
- [source 1]
- [source 2]

---

Good research queries

Asset deep dives:

• "Investment thesis for [TOKEN]: fundamentals, competition, risks, and outlook"
• "On-chain health of [TOKEN]: activity, retention, revenue"

Sector analysis:

• "Competitive landscape of Ethereum L2 rollups in 2025"
• "State of RWA tokenization: market size, key players, regulatory risks"
• "DeFi lending protocols: risk comparison across Aave, Compound, and Morpho"

Narrative research:

• "What is the AI x crypto narrative and which projects are best positioned?"
• "Is the memecoin supercycle over? Evidence and counterarguments"

Macro:

• "How does the current macro environment affect crypto market structure?"

---

Notes

• Deep research is async — start the job, then come back; do not busy-wait
• Report quality is significantly higher than single-shot AI chat
• If polling for >20 minutes with processing status, the job may be stuck — cancel and retry
• Payments in USDC on Base (--chain base)
• Free polling: Step 2 GET requests do not trigger payment

Related skills

• messari-x402 — Core endpoint reference and quick AI chat (/ai/v2/chat/completions)
• messari-token-research — Faster, cheaper token research using multiple data endpoints
• messari-alpha-scout — Surface topics worth deep-researching
• messari-funding-intel — Complement with funding data