nyne-deep-research（尼恩深入研究）

Nyne Deep Research Skill

Research any person by email, phone, social URL, or name. Returns a comprehensive intelligence dossier with psychographic profile, social graph, career analysis, conversation starters, and approach strategy.

Important: This API is async with high latency (2-5 min processing). You submit a request, get a request_id, then poll until complete.

Agent Instructions

When presenting results to the user, show maximum depth — display every section of the dossier in full. Do not summarize or truncate. Walk through each section sequentially:

1. Identity Snapshot — who they are
2. Career DNA — trajectory and superpower
3. Psychographic Profile — values, motivations, archetypes
4. Personal Life & Hobbies — interests outside work
5. Social Graph Analysis — inner circle, professional network, interest graph
6. Interest Cluster Deep Dive — all 9 clusters with detail
7. Content & Voice Analysis — topics, tone, opinions, quotes
8. Key Relationships — full list with relationship nature and importance
9. Conversation Starters — all 4 hook categories
10. Recommendations / How Others See Them — reputation signals
11. Warnings & Landmines — topics to avoid, sensitivities
12. Creepy-Good Insights — non-obvious findings with evidence
13. Approach Strategy — best angle, topics, what not to do

Also include enrichment data (contact info, work history, education) and note whether following and articles sections have data.

If dossier is null, present the enrichment data and let the user know the full dossier was not generated (this can happen with duplicate or cached requests — resubmit with a different identifier).

If following or articles are null, simply note they were not available for this person.

Setup

Required environment variables:

• NYNE_API_KEY — your Nyne API key
• NYNE_API_SECRET — your Nyne API secret

Get credentials at https://api.nyne.ai.

Set these in your shell before running any commands:

export NYNE_API_KEY="your-api-key"
export NYNE_API_SECRET="your-api-secret"

To persist across sessions, add the exports to your shell profile (~/.zshrc, ~/.bashrc, etc.) or create a .env file and source it:

# Create .env file (keep out of version control)
echo 'export NYNE_API_KEY="your-api-key"' >> ~/.nyne_env
echo 'export NYNE_API_SECRET="your-api-secret"' >> ~/.nyne_env
source ~/.nyne_env

Verify they're set:

echo "Key: ${NYNE_API_KEY:0:8}... Secret: ${NYNE_API_SECRET:0:6}..."

Important: JSON Handling

The API response can contain control characters in JSON string values that break jq. All examples below use a nyne_parse helper that pipes through python3 to clean and re-encode the JSON before passing to jq. Define it once per session:

nyne_parse() {
  python3 -c "
import sys, json, re
raw = sys.stdin.read()
clean = re.sub(r'[\-\]+', ' ', raw)
data = json.loads(clean)
json.dump(data, sys.stdout)
"
}

Quick Start

Submit a research request by email and poll until complete:

# Define helper (strips control chars, re-encodes clean JSON)
nyne_parse() {
  python3 -c "
import sys, json, re
raw = sys.stdin.read()
clean = re.sub(r'[\-\]+', ' ', raw)
data = json.loads(clean)
json.dump(data, sys.stdout)
"
}

# Submit research request
REQUEST_ID=$(curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"email": "someone@example.com"}' | nyne_parse | jq -r '.data.request_id')

echo "Request submitted: $REQUEST_ID"

# Poll until complete (checks every 5s, times out after 10 min)
SECONDS_WAITED=0
while [ $SECONDS_WAITED -lt 600 ]; do
  curl -s "https://api.nyne.ai/person/deep-research?request_id=$REQUEST_ID" \
    -H "X-API-Key: $NYNE_API_KEY" \
    -H "X-API-Secret: $NYNE_API_SECRET" | nyne_parse > /tmp/nyne_response.json
  STATUS=$(jq -r '.data.status' /tmp/nyne_response.json)
  echo "Status: $STATUS ($SECONDS_WAITED seconds elapsed)"
  if [ "$STATUS" = "completed" ]; then
    jq '.data.result' /tmp/nyne_response.json
    break
  elif [ "$STATUS" = "failed" ]; then
    echo "Research failed."
    jq . /tmp/nyne_response.json
    break
  fi
  sleep 5
  SECONDS_WAITED=$((SECONDS_WAITED + 5))
done

if [ $SECONDS_WAITED -ge 600 ]; then
  echo "Timed out after 10 minutes. Try polling manually with request_id: $REQUEST_ID"
fi

Submit Research (POST)

Endpoint: POST https://api.nyne.ai/person/deep-research

Headers:

Content-Type: application/json
X-API-Key: $NYNE_API_KEY
X-API-Secret: $NYNE_API_SECRET

Parameters:

At least one identifier is required: email, phone, social_media_url, or name with company/city.

Examples

By email:

curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"email": "someone@example.com"}' | nyne_parse | jq .

By social media URL:

curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"social_media_url": "https://twitter.com/elonmusk"}' | nyne_parse | jq .

By multiple social URLs:

curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"social_media_url": ["https://twitter.com/elonmusk", "https://linkedin.com/in/elonmusk"]}' | nyne_parse | jq .

By name + company:

curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"name": "Jane Smith", "company": "Acme Corp"}' | nyne_parse | jq .

By phone:

curl -s -X POST "https://api.nyne.ai/person/deep-research" \
  -H "Content-Type: application/json" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" \
  -d '{"phone": "+14155551234"}' | nyne_parse | jq .

Submit response (HTTP 202):

{
  "success": true,
  "data": {
    "request_id": "abc123-def456-...",
    "status": "pending"
  }
}

Poll for Results (GET)

Endpoint: GET https://api.nyne.ai/person/deep-research?request_id={id}

Headers: Same X-API-Key and X-API-Secret as above.

Status progression: pending → enriching → gathering → analyzing → completed (or failed)

This typically takes 2-3 minutes.

Check status once

curl -s "https://api.nyne.ai/person/deep-research?request_id=$REQUEST_ID" \
  -H "X-API-Key: $NYNE_API_KEY" \
  -H "X-API-Secret: $NYNE_API_SECRET" | nyne_parse | jq '{status: .data.status, completed: .data.completed}'

Full polling loop

SECONDS_WAITED=0
TIMEOUT=600  # 10 minutes

while [ $SECONDS_WAITED -lt $TIMEOUT ]; do
  curl -s "https://api.nyne.ai/person/deep-research?request_id=$REQUEST_ID" \
    -H "X-API-Key: $NYNE_API_KEY" \
    -H "X-API-Secret: $NYNE_API_SECRET" | nyne_parse > /tmp/nyne_response.json
  STATUS=$(jq -r '.data.status' /tmp/nyne_response.json)

  echo "[$(date +%H:%M:%S)] Status: $STATUS ($SECONDS_WAITED s)"

  case "$STATUS" in
    completed)
      jq '.data.result' /tmp/nyne_response.json
      break
      ;;
    failed)
      echo "Research failed:"
      jq '.data' /tmp/nyne_response.json
      break
      ;;
    *)
      sleep 5
      SECONDS_WAITED=$((SECONDS_WAITED + 5))
      ;;
  esac
done

if [ "$STATUS" != "completed" ] && [ "$STATUS" != "failed" ]; then
  echo "Timed out. Resume polling with request_id: $REQUEST_ID"
fi

Response Structure

When status is completed, the response looks like:

{
  "success": true,
  "timestamp": "2025-01-15T12:00:00Z",
  "data": {
    "status": "completed",
    "completed": true,
    "request_id": "abc123-...",
    "created_on": "2025-01-15T11:57:00Z",
    "completed_on": "2025-01-15T12:00:00Z",
    "result": {
      "enrichment": { ... },
      "dossier": { ... },
      "following": { ... },
      "articles": [ ... ]
    }
  }
}

result sections

Dossier Sections Reference

The dossier object contains 15 sections:

identity_snapshot

Top-level identity summary.

• full_name, current_role, company, location, age_estimate, emails, social_profiles, headline, birthday, self_description

career_dna

Career trajectory and strengths.

• trajectory, superpower

psychographic_profile

Values, motivations, and personality archetypes.

• values, motivations, archetypes, political_leanings, cluster_analysis

personal_life_hobbies

Interests and personality outside of work.

• life_outside_work, entertainment_culture, personal_passions, active_hobbies_sports, quirks_personality

social_graph_analysis

Network and relationship mapping.

• inner_circle, professional_network, personal_interest_graph

interest_cluster_deep_dive

Deep analysis across 9 interest clusters.

• tech, sports_fitness, music_entertainment, food_lifestyle, causes_politics, intellectual_interests, geographic_ties, personal_network, unexpected_surprising

content_voice_analysis

How they communicate and what they care about.

• topics, tone, humor_style, strong_opinions, frustrations, notable_quotes, recent_wins

content_analysis

Alias of content_voice_analysis (kept for backward compatibility).

key_relationships

List of ~25 objects describing important connections.

• Each: name, handle, followers, relationship_nature, why_important

key_influencers

Alias of key_relationships.

conversation_starters

Hooks to open conversation, in 4 categories.

• professional_hooks, personal_interest_hooks, current_events_hooks, shared_experience_hooks

recommendations_how_others_see_them

Public perception and reputation signals.

• highlighted_qualities, colleague_descriptions, patterns_in_praise

warnings_landmines

Topics and areas to avoid.

• topics_to_avoid, political_hot_buttons, sensitive_career_history, competitors_they_dislike

creepy_good_insights

Non-obvious insights derived from data.

• List of objects: insight, evidence

approach_strategy

How to approach this person.

• best_angle, topics_that_resonate, personal_interests_to_reference, shared_connections, what_not_to_do

Useful jq Filters

After polling completes, the clean response is at /tmp/nyne_response.json. Use jq directly on the file:

# Extract identity snapshot
jq '.data.result.dossier.identity_snapshot' /tmp/nyne_response.json

# Get all conversation starters
jq '.data.result.dossier.conversation_starters' /tmp/nyne_response.json

# List key relationships (name + why important)
jq '.data.result.dossier.key_relationships[] | {name, why_important}' /tmp/nyne_response.json

# Get approach strategy
jq '.data.result.dossier.approach_strategy' /tmp/nyne_response.json

# Extract a specific dossier section by name
jq --arg s "psychographic_profile" '.data.result.dossier[$s]' /tmp/nyne_response.json

# Get enrichment contact info
jq '.data.result.enrichment | {emails, phones, linkedin_url, twitter_url}' /tmp/nyne_response.json

# Check processing status only
jq '{status: .data.status, completed: .data.completed, request_id: .data.request_id}' /tmp/nyne_response.json

# Get warnings and landmines
jq '.data.result.dossier.warnings_landmines' /tmp/nyne_response.json

# List all creepy-good insights
jq '.data.result.dossier.creepy_good_insights[].insight' /tmp/nyne_response.json

Error Handling

Rate Limits & Costs

• Rate limits: 10 requests/minute, 100 requests/hour
• Cost: 100 credits per research request
• Processing time: Typically 2-3 minutes, up to 5 minutes