Data Guide

This guide covers the concepts and patterns you need to work effectively with AIXBT data, regardless of which surface you use (CLI, MCP, or REST API).

Core Concepts

Momentum

spikingScore measures social breakout strength: recent acceleration in community breadth, author breadth, and short-window chatter, penalized when attention is already saturated or concentrated in one source. It is not raw mention volume.

Project responses include momentumContext (when selected) which provides pre-computed stability context: a trajectory label (rising/stable/declining/spike), cluster count, and leaderboard presence data. The leaderboard.top10 and leaderboard.top25 sub-objects each carry a stabilityLabel — flash (<2h), fresh (2-6h), persistent (6-24h), or durable (>=24h) — indicating how long the project has continuously held that position. The leaderboard is volatile: most projects hold top-10 for under 2 hours, so flash positions are the norm. Persistent and durable positions indicate genuine sustained attention.

For climbing-growth rankings — which projects are accumulating attention over multiple recent buckets without a single concentrated burst — use sortBy=climbingScore, the top-projects-climbing action, or GET /v2/projects/top/climbing.

Active

activeScore is the number of hours in the last 24 that had at least one mention (0-24). It measures recent activity coverage, not sustained growth. A project can have high active coverage but low spiking momentum (active but not accelerating) or high spiking momentum but moderate active coverage (rapidly emerging from a smaller base).

Climbing Growth

climbingScore measures attention accumulation over the last 72h using 4h buckets. It rewards final-day growth, recent 12h/8h follow-through, distinct author and cluster breadth, and penalizes activity concentrated in one 4h bucket or one author. Use it for "slowly growing" or "steadily gaining attention" questions.

Intel

Structured events extracted from community discussion. Each intel item has a category, a description, timestamps, and source clusters. Intel is deduplicated: if multiple sources report the same event, they reinforce a single intel item rather than creating duplicates.

Key fields:

• detectedAt -- When the intel was first spotted
• reinforcedAt -- When it was last reinforced by new evidence
• observationCount -- Number of source observations supporting this intel item (use this instead of counting activity entries)
• activity -- Optional timeline of intel evolution, controlled by activity=all|observations|none. Each entry has date, source, clusters, and what the new evidence said
• hasOfficialSource -- True when a source observation came from an official project account

Intel Categories

Clusters

Community clusters are overlapping account archetypes in the crypto conversation. Cross-cluster breadth carries more weight than single-cluster volume because it shows chatter moving beyond one archetype. Use cluster counts as coverage signals, then check source and reinforcement details for confirmation.

Metrics

Market data available on project objects without an external lookup: price, market cap, 24h volume, 24h change, 24h high/low, and 7d/14d/30d price change.

Grounding

Structured market context covering crypto and tradfi. Updated hourly. Use it to contextualize intel and momentum data against broader market conditions.

Query Patterns

These describe analysis workflows. Map to your surface's commands or tools.

Research a project

Look up by name, ticker, or contract address. Get full details. Check momentum history. Scan recent intel.

Assess: is momentum rising or declining? Which clusters are driving it? What intel categories dominate? Heavy TECH_EVENT intel means development progress. Heavy OPINION_SPECULATION means narrative without substance.

Note: project detail responses include the 10 most recent intel items. Use the intel endpoint directly for a fuller history.

Market overview

Fetch top projects by spiking momentum (what's spreading fastest), active coverage (what has recent active-hour coverage), or climbing growth (what is accumulating across 4h buckets). Fetch recent intel across all projects. Cross-reference: which projects appear in multiple rankings?

To identify narratives, fetch a larger set and synthesize across the results -- group projects that share intel themes or cluster overlap.

Climbing leaders

Use top-projects-climbing or sortBy=climbingScore to find projects that are growing more gradually across the last 72h. Compare against the spiking list to distinguish what's spiking right now from what is steadily accumulating attention.

Cluster-Specific

List clusters. Identify relevant ones by name and description. Fetch intel from those clusters. Compare against the broader momentum list to find what a group noticed that the market hasn't.

Category scan

Filter intel by category. Combine strategically:

Time-based queries

• Use detectedAfter/detectedBefore to find intel by when it was first spotted
• Use reinforcedAfter/reinforcedBefore to find intel actively discussed in a period
• Use the momentum endpoint with start/end for how attention evolved over time

These can be combined: "intel detected in the last month that is still being reinforced today" uses both date filter types.

Interpreting Patterns

High momentum, sparse intel -- Talked about but few verifiable events. Could be hype, speculation, or early-stage buzz before developments are announced. Treat with caution.

Heavily reinforced intel across clusters -- Strong narrative with substance. The intel is reinforced by multiple sources and spans several community clusters. Stronger pattern when the underlying intel is substantive.

High momentum, single cluster -- Attention remains concentrated in one community cluster. Could be organic early discovery or coordinated promotion.

Flash stability label despite high momentum -- The project entered the top list recently (<2h) and may not sustain. High momentum with a flash label is common and should not be treated as notable. Wait for persistent or durable labels before treating the position as established.

Momentum declining, intel still active -- The news is real but attention is fading. The story may be priced in or losing novelty.

Old intel, recent reinforcement -- An event from days or weeks ago is still being discussed. Often means the implications are still unfolding.

No intel, but project is tracked -- Chatter exists but hasn't produced verifiable events. Common for meme tokens and speculation-driven projects.

Cluster progression in activity log -- Track which cluster reported first and how it spread. Early clusters often have better intel quality for that domain.

Cross-Referencing with External Data

AIXBT includes identifiers for looking up projects in other systems:

• xHandle -- The project's X/Twitter handle
• coingeckoData.apiId -- For historical price data and deeper market context
• tokens[].address + tokens[].chain -- Contract addresses for on-chain lookups, DEX data, or block explorers
• coingeckoData.categories -- Sector tags for grouping projects by theme