Agent Skill Reference

This guide teaches you how to think with AIXBT data. It tells you when to call what, in what order, and how to interpret what comes back.

What AIXBT Data Tells You

AIXBT is a real-time market intelligence platform for crypto. It groups X accounts into clusters using social graph analysis, then extracts structured signals from their activity.

What you get:

• Which projects are gaining attention across independent communities
• What concrete events are happening — launches, partnerships, funding, whale moves, risk alerts
• How narratives form and spread — which clusters spotted something first, how it propagated
• Whether attention is real or noise — cross-cluster convergence vs single-community hype

Key Fields

When you receive a project object, here's what the fields mean:

momentumScore — Rate at which new clusters are picking up this project. This is not mention volume. A project discussed by 15 independent clusters scores higher than one with more total mentions confined to 2-3 groups. Momentum captures narrative spread, not loudness.

popularityScore — Number of hours in the last 24 that had at least one mention (0-24). High popularity means sustained, consistent attention. High momentum means accelerating spread to new audiences. A project can have high popularity but low momentum (established, stable attention) or high momentum but moderate popularity (rapidly emerging but still early).

signals — Structured events extracted from community discussion. Each signal has a category, a description, timestamps, and a list of clusters that reported it. Categories range from concrete events (TECH_EVENT, PARTNERSHIP) to community sentiment (OPINION_SPECULATION). Signals are deduplicated: if multiple sources report the same event, they reinforce a single signal rather than creating duplicates.

reinforcedAt vs detectedAt — detectedAt is when the signal was first spotted. reinforcedAt is when it was last reinforced by new evidence. A reinforced signal is an ongoing topic still in the zeitgeist, spreading across clusters. A newly detected signal is early in its narrative. Neither is more important — they represent different stages. Use both timestamps to understand where a signal is in its lifecycle.

clusters (on signals) — Which communities reported this event. Signals seen by multiple clusters carry more weight than single-cluster signals.

activity (on signals) — The reinforcement history. Each entry has date, source (platform provenance), cluster (which community reported it), incoming (what the new evidence said), and result (the merged signal description after incorporating it). Track these to see how a signal evolved and which platforms/clusters contributed over time.

metrics — Current market data: usd (price), usdMarketCap, usd24hVol, usd24hChange (24h percentage), lastUpdatedAt.

coingeckoData — Project identity from CoinGecko: apiId, symbol, categories, description, homepage. This is metadata for identification and external lookups.

tokens — Contract addresses by blockchain. Each entry has chain, address, and source (where the data came from). Use these with block explorers, DEX aggregators, or on-chain analytics tools.

Query Patterns

Research a specific project

When the user asks about a named project ("What's happening with Ethereum?", "Tell me about Hyperliquid"):

1. GET https://api.aixbt.tech/v2/projects?names=ethereum&limit=5
   → Find the project, get its ID, signals, momentum, and metadata

2. GET https://api.aixbt.tech/v2/projects/{id}
   → Full project details (useful when you already have an ID from a signal or other reference)

3. GET https://api.aixbt.tech/v2/projects/{id}/momentum
   → Hourly momentum history with cluster breakdown (defaults to last 7 days; pass start/end for longer)

4. GET https://api.aixbt.tech/v2/signals?projectIds={id}&limit=50
   → Full signal history — projects embed only their 10 most recent signals, use this for more

Not every step is always needed. If the list response gives you enough signals and the momentum score tells the story, stop there. Use the detail and momentum endpoints when you need depth.

When analyzing: check momentum trend (rising, stable, or declining), which clusters are driving attention, and what signal categories dominate. Heavy TECH_EVENT signals mean development progress. Heavy OPINION_SPECULATION means narrative without substance.

What's trending / market overview

When the user asks broad questions ("What's hot?", "What should I pay attention to?", "Daily overview"):

1. GET https://api.aixbt.tech/v2/projects?limit=20&sortBy=momentumScore
   → Top projects by momentum — what's spreading fastest across communities

2. GET https://api.aixbt.tech/v2/signals?limit=50
   → Recent signals across all projects — what events are happening right now

3. Cross-reference: Which projects appear in both? What signal categories dominate?

If the user wants established attention rather than emerging trends, use sortBy=popularityScore instead.

To identify narratives, fetch a larger set (e.g. limit=50) and synthesize across the results. A prompt like this works well:

Given these projects and their signals, identify the dominant narratives this week. Group projects that share signal themes or cluster overlap. What stories connect them? What's noise?

Community-specific queries

When the user asks about specific groups ("What are VCs talking about?", "What's trending among traders?"):

1. GET https://api.aixbt.tech/v2/clusters
   → Get all cluster names and descriptions

2. Identify relevant clusters by name
   → Select ALL that match (e.g., for "traders": trader-1, trader-2, crypto-traders, etc.)

3. GET https://api.aixbt.tech/v2/signals?clusterIds={id1},{id2},{id3}&reinforcedAfter={48h_ago}
   → Signals from those communities specifically

The clusters endpoint returns both names and descriptions. Read the descriptions to understand what each community represents before selecting.

Cross-reference community-specific signals against the top momentum list to surface what a group has noticed that the broader market hasn't.

Category-specific analysis

When the user is interested in specific types of events ("Any whale moves?", "What partnerships were announced?"):

GET https://api.aixbt.tech/v2/signals?categories=WHALE_ACTIVITY,MARKET_ACTIVITY&reinforcedAfter={7d_ago}&limit=50

Categories can be combined strategically:

For risk categories, check whether signals are being reinforced (growing concern) or are isolated single-mentions. Cross-reference with trending projects to flag popular tokens carrying risk signals.

Time-based queries

For historical analysis ("What happened this week?", "How has sentiment changed since the crash?"):

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

These can be combined. "Signals detected in the last month that are still being reinforced today" requires both date filter types.

Comparative analysis

When comparing projects ("Compare momentum between Project A and Project B"):

1. GET https://api.aixbt.tech/v2/projects?names=projectA,projectB
   → Both projects in one call (names filter uses OR logic within the parameter)

2. GET https://api.aixbt.tech/v2/projects/{idA}/momentum  +  GET https://api.aixbt.tech/v2/projects/{idB}/momentum
   → Side-by-side momentum history

3. Compare: cluster overlap, momentum trajectory, signal categories

Pagination

All list endpoints return paginated results. The response includes page, limit, totalCount, and hasMore. Default and maximum limit is 50. Use page to fetch additional results when hasMore is true.

Signal Categories

Interpreting Patterns

High momentum, few signals — The project is getting talked about, but concrete events are sparse. Could be hype, speculation, or early-stage buzz before developments are announced. Treat with caution.

Many reinforced signals across clusters — Strong narrative with substance. Multiple independent communities are confirming the same events. This is the highest-conviction pattern.

High momentum, single cluster — One community is excited, but it hasn't spread. Could be organic early discovery or coordinated promotion. Check whether the cluster is known for quality signal.

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

Old signal, recent reinforcement — An event from days or weeks ago is still being discussed. Often means the implications are still unfolding or the event is larger than initially understood.

No signals, but project is tracked — Activity exists (it contributes to momentum) but hasn't produced verifiable events. Common for meme tokens and speculation-driven projects.

Signal activity log shows cluster progression — Track which cluster reported first and how it spread. Early clusters often have better signal quality for that domain. If trading-focused clusters spotted it before general audiences, the market implications may be stronger.

Cross-Referencing with External Data

AIXBT includes identifiers you can use to look up projects in other systems:

• xHandle — The project's X handle, for linking to or searching their profile
• coingeckoData.apiId — Identifier for looking up historical price data and deeper market context beyond what metrics provides
• tokens[].address + tokens[].chain — Contract addresses for on-chain lookups, DEX data, or block explorer queries
• coingeckoData.categories — Sector tags useful for grouping projects by theme

Further Reading

• API Reference — Full endpoint documentation with parameters and response schemas
• Documentation Index — Index of all AIXBT documentation