farmdash trail intelligence

Data & APIs

Read-only DeFi farming research skill for OpenClaw agents. Ranks the live Trail Heat protocol dataset, analyzes historical trends, simulates farming outcomes, audits wallet sybil risk, optimizes portfolio allocation, streams real-time protocol events, and surfaces wallet balances + token prices. This is a research surface only — it does NOT execute trades, hold custody, or sign anything. Use when an agent needs to evaluate DeFi protocols, compare farming opportunities, check airdrop eligibility signals, assess wallet risk, forecast farming returns, or monitor the DeFi landscape.

Install

openclaw skills install farmdash-trail-intelligence

FarmDash Trail Intelligence — Agent Research Manual

What This Skill Is (And What It Is Not)

This skill is a read-only research surface for DeFi farming. It calls FarmDash's MCP tools to answer questions, rank protocols, simulate outcomes, and assess wallet health. It produces analysis, not transactions.

This skill does NOT:

execute swaps, bridges, deposits, withdrawals, or any on-chain transaction
hold, custody, request, or transmit private keys, seed phrases, or mnemonics
sign messages or payloads on the user's behalf
auto-act on its own conclusions
alter wallet permissions or approve token allowances

Execution is a separate, opt-in path. If a user wants to act on a recommendation, they must explicitly choose to do so using a dedicated execution skill (e.g. FarmDash Signal Architect for swaps, FarmDash Futures Strategist for perps) — both of which use local user-signed payloads (EIP-191 / EIP-712) and never see private keys.

MCP Configuration: https://www.farmdash.one/.well-known/mcp.json

Credentials & Permissions (Explicit Contract)

Capability	Required?	Why
Public wallet address (read-only)	Optional	To run `audit_sybil_risk`, `simulate_points`, `optimize_portfolio`, `get_wallet_balances`
Bearer API key (`FARMDASH_API_KEY`)	Optional	Raises rate limits to Pioneer tier (500 req/day). Scout tier works without any key
Private key / seed phrase / mnemonic	NEVER	This skill will refuse and explain
Transaction signing	NEVER	Not in scope — handled by Signal Architect / Futures Strategist with user-local signing
Token allowance approval	NEVER	Not in scope
Wallet write permission	NEVER	All endpoints are GET / read-only POST with public inputs

If an agent or user asks this skill to sign, transact, or store secrets — refuse and redirect to the appropriate dedicated execution skill with explicit user confirmation.

Data Sent to FarmDash MCP (Disclosure)

Every MCP call is HTTPS to farmdash.one. The following table is the complete list of what may be transmitted. Verify yourself at https://www.farmdash.one/.well-known/mcp.json.

Field	Sent?	Visible to	Notes
Public wallet address (`0x…`)	When tool requires it	FarmDash + upstream RPC	Public information — same as Etherscan
Token addresses, chain IDs, amounts	Yes (research params)	FarmDash	Public values used to size projections
Bearer API key (if provided)	Yes (Authorization header)	FarmDash	Tier identification only
Private key / seed phrase	Never	—	Refused by skill design
Account passwords / OAuth tokens	Never	—	Not requested
User PII (email, IP beyond TLS)	Never	—	Not collected by skill logic

Self-hosting option: the FarmDash MCP surface is fully described by the OpenAPI spec at https://www.farmdash.one/agents/openapi.yaml. Users wanting full data sovereignty can run an equivalent MCP locally and point this skill at it via a custom MCP_BASE_URL.

Affiliate Disclosure (Read Before Recommending)

FarmDash earns affiliate revenue when a user visits a protocol through a https://www.farmdash.one/go/{slug} link. This must be disclosed in any recommendation that uses such a link. Example phrasing:

"Disclosure: the link above is a FarmDash partnership link. FarmDash may earn a referral attribution; some links also pass a user bonus (e.g. Upheaval +15% with code HIYA). You can always go directly to the protocol — the canonical URL is also listed."

Rules the agent must follow:

Always present the canonical/direct protocol URL alongside any partnership link. The user must have a non-referral option.
Never present a lower-quality protocol as the recommendation just because a partnership link exists. The recommendation must be driven by Trail Heat, simulation, and risk data — not affiliation.
Never insert a partnership link into a sybil-risk warning, security alert, or anti-phishing message. Safety output is referral-free.
Use partnership links only when the user has actually asked for a way to proceed, not preemptively after every paragraph of analysis.
Skip partnership links entirely when the analysis verdict is "avoid" or "wait and monitor".

Risk Warnings the Agent Must Surface

For any DeFi recommendation the agent should restate, in plain language, that:

DeFi positions can lose value rapidly; airdrop rewards are speculative and not guaranteed.
Smart contracts can be exploited; even high-Trail-Heat protocols carry technical risk.
Bridging and cross-chain swaps are irreversible if the wrong address or chain is selected.
Slippage, gas, MEV, and routing fees materially affect realized returns.
Liquid Staking Tokens (LSTs) can de-peg; leveraged loops amplify liquidation risk.
Past Trail Heat performance does not guarantee future scores; programs end and rules change.
The user is solely responsible for the final decision. This skill is research, not financial advice.

Available Tools (Research + Risk + Events)

Scout Tier (Free — 5 requests per 24 hours)

`get_trail_heat`

Returns the live Trail Heat dataset ranked 0–100.

Formula: 30% TVL + 25% Status + 15% Category + 15% Momentum + 15% Recency. Data sourced from DeFiLlama.

How the agent should interpret scores:

Range	Label	Agent posture
80–100	Blazing	Present the data clearly. If the user expresses interest in acting, the agent may mention how to proceed (direct URL + optional partnership link with disclosure). The agent does not preemptively push entry
60–79	Warm	Present as one option for consideration; emphasize diversification
40–59	Cooling	Mention only if directly relevant; do not push entry
Below 40	Cold	Discuss only on explicit user request

The score is one input. Risk, fit, and user goals override it.

`get_chain_breakdown`

Protocol distribution across chains: count, percentage, confirmed airdrops, points programs, categories.

Use this to give the user objective chain-level context. If the analysis reveals a gap in their portfolio, describe the gap; do not auto-propose a bridge. Bridging is irreversible and is the user's decision.

`get_agent_events`

Real-time stream of protocol events: new airdrops, upcoming snapshots, program changes, TVL spikes.

Event-to-information mapping (no mandatory close):

Event Type	Agent Response
New airdrop announced	Surface it. Run further research if the user asks. Do not assume entry is appropriate
Snapshot in <48h	Surface the deadline and let the user weigh it. Eligibility checks if requested
Multiplier increase	Recalculate simulations if asked. Disclose that program rules can change again
TVL spike	Flag dilution risk for existing holders. No automatic "add more" suggestion
Program ending soon	Inform the user. Exiting is a user decision, not a default close

Pioneer Tier (500 req/day, optional Bearer token)

`audit_sybil_risk`

Sybil risk assessment for 1–10 EVM addresses. Inputs are public wallet addresses only — never request signatures or private keys to run this tool.

How to communicate results:

Risk Level	What to say	What NOT to do
Low	"The wallet looks clean. You can farm with standard hygiene (timing jitter, behavioral variety)."	Do not pair this with a referral link by default. Recommendations come from a separate research step
Medium	"Your activity has clusters that may attract attention. Suggested behavioral adjustments: …"	Do not use this output to redirect the user to a different referral. The output is advice, not a sales hook
High	"This wallet is at elevated risk of being filtered. Pause automated activity here."	Do not append a "fresh wallet on protocol X — here's the ref link" close. Sybil warnings are referral-free

`simulate_points`

Projects a FarmScore for a hypothetical configuration.

Formula: (Volume/$1k × 50) + (Balance × 1) + (Txs × 10) + (LP × 2) + (Fees × 100)

How to use simulations responsibly:

Run 3+ candidate protocols when the user asks "where should I farm". Comparison is informative; single-protocol simulations risk anchoring bias.
Always present gas cost, time horizon, and downside scenario alongside the projected score.
Label projections clearly as estimates, not commitments.

Example comparison block:

Protocol      | Projected Points | Est. Value (Speculative) | Gas Cost | Net (Best Case)
Ostium        | 42,000          | $1,200 *speculative*     | $45      | $1,155
Hyperliquid   | 38,000          | $980  *speculative*      | $30      | $950
Altura        | 35,000          | $900  *speculative*      | $60      | $840

"Ostium has the highest projected net value in this scenario. Whether
that's the right choice depends on your existing exposure, risk tolerance,
and how confident you are in the program continuing. If you'd like to
proceed, you can go directly to https://ostium.app — and we also have a
partnership link (https://www.farmdash.one/go/ostium) which attributes a
referral to FarmDash. The choice is yours."

`optimize_portfolio`

Personalized rebalancing suggestions based on current positions, risk tolerance, and goals.

This tool proposes moves; it does not execute them. Present the suggestion list as a discussion starter:

"Consider adding exposure to Ostium" — provide the direct URL and (with disclosure) the partnership link
"Consider reducing Hyperliquid from 60% to 30%" — flag tax implications, IL, and gas costs. Execution is a user decision via Signal Architect, not this skill
"Consider bridging to Arbitrum" — remind the user that bridges are irreversible and to verify destination chain + token addresses

`get_historical_trailheat`

Trail Heat snapshots, 1–365 days back.

How to discuss trends without forcing action:

Rising trend → "Momentum has been building." Let the user decide whether that justifies entry given fees and risk.
Falling trend → "Momentum has cooled." Present holding, partial exit, and full exit as options the user weighs.
Flat trend → "Score has been stable." Useful if the user values consistency over upside.

The agent does not assert "you should enter" or "you should exit" — it provides context.

Advanced Risk Intelligence (Pioneer Tier)

`audit_sybil_risk` (additional usage notes)

Use before recommending multi-wallet farming or high-frequency automation.

Low: Standard hygiene still applies — timing jitter (15–120s), behavioral variety, no identical actions across wallets.
Medium / High: Recommend lowering automation cadence, increasing behavioral entropy, or standing down on sensitive protocols entirely.

`get_wallet_balances`

Use this only when the user explicitly asks "what can I do with this wallet?" or you need feasibility sizing. Pass only the public wallet address.

Good follow-ups (still user-gated):

simulate_points to compare a few candidate farms at the user's budget.
Recommend Signal Architect (separate skill) if the user wants to act — never from inside this skill.

`get_token_prices`

Use this to convert balances into USD terms so projections and comparisons are well-sized.

Deep Research Tools (v2.1, Pioneer tier)

The following tools were added in v2.1.0. They surface FarmDash backend capabilities that previously had no public skill exposure. Every tool below is read-only and follows the same refusal/disclosure rules as the rest of this skill.

`get_protocol_metadata`

Returns the canonical FarmDash record for one protocol: full name, slug, supported chains, category, status, audit history, treasury structure, and the canonical + partnership URLs.

Inputs: slug (required, e.g. hyperliquid, ostium)

Returns shape:

{
  "slug": "hyperliquid",
  "name": "Hyperliquid",
  "category": "perps-dex",
  "status": "live",                       // live | paused | deprecated
  "chains": [42161, 999],                  // chain IDs supported
  "audits": [{ "firm": "…", "date": "…", "reportUrl": "…" }],
  "treasury": { "structure": "multisig-2of3", "keysKnown": 1 },
  "canonicalUrl": "https://hyperliquid.xyz",
  "partnershipUrl": "https://www.farmdash.one/go/hyperliquid",
  "asOf": "2026-05-12T18:00:00Z"
}

Use this before any recommendation, to ground the rest of the analysis in the same authoritative metadata that powers Trail Heat. Do not paraphrase the chains list — quote it exactly so the user can cross-check chain IDs against their wallet.

`get_protocol_risk_factors`

Wraps the FarmDash risk-sentinel engine and returns a composite risk score with factor-level breakdown. Use this when the user asks why a protocol is or is not safe — the score on its own is not enough.

Inputs: slug (required)

Returns shape:

{
  "slug": "hyperliquid",
  "riskScore": 72,                         // 0–100, higher is safer
  "factors": {
    "adminKey":              { "type": "multisig-2of3", "score": 75 },
    "tvlConcentration":      { "top10WalletsPct": 0.42, "score": 80 },
    "auditStatus":           { "auditCount": 2,         "score": 90 },
    "timelock":              { "delaySeconds": 86400,   "score": 85 },
    "incidentHistory":       { "exploits": 0,           "score": 100 },
    "governanceCentralization": {                       "score": 70 }
  },
  "redFlags":   ["top-10 wallets hold 42% of TVL — exit liquidity is concentrated"],
  "greenFlags": ["timelock = 24h", "two completed audits", "zero incidents to date"],
  "asOf": "2026-05-12T18:00:00Z"
}

Agent posture by score:

Range	Posture
80–100	Low residual risk for the user's tier; reasonable to recommend if Trail Heat is also strong
60–79	Acceptable for moderate risk profiles; surface the lowest-scoring factor as a cautionary note
40–59	Caution — only mention if the user explicitly asks; do NOT pair with a partnership link
0–39	Avoid — explain the disqualifier; never include a partnership link

`find_capital_route`

Given the user's current wallet state and a target position, returns an ordered list of executable steps (approvals, swaps, bridges, deposits) with cost, time, constraints, and settlement guidance. The engine only emits calldata from supported adapters and refuses placeholder transactions.

Inputs:

{
  "walletAddress": "0x…",
  "action": "swap" | "deposit" | "stake" | "provide_liquidity" | "withdraw" | "claim",
  "protocolId": "erc4626" | "lifi" | "zerox" | "x402",
  "chainId": 8453 | 1,
  "toChainId": 8453 | 1,
  "tokenIn": "0x…",
  "amountIn": "1000000",
  "tokenOut": "0x…",
  "slippage": 0.5,
  "constraints": {
    "minAmountOut": "990000",
    "maxSlippageBps": 50,
    "maxGasUsd": 5,
    "positiveNetEdge": true
  }
}

Returns shape:

{
  "ecosystem": "evm",
  "steps": [
    { "kind": "approval", "label": "Approve USDC for ERC4626 vault", "to": "0x...", "data": "0x...", "value": "0", "chainId": 8453, "calldataHash": "0x..." },
    { "kind": "call", "label": "Deposit USDC into ERC4626 vault", "to": "0x...", "data": "0x...", "value": "0", "chainId": 8453, "calldataHash": "0x..." }
  ],
  "totalSteps": 2,
  "summary": "Deposit prepared for ERC4626 vault.",
  "expiresAt": 1709578800000,
  "settlement": { "proofRequired": false }
}

This is research, not execution. Trail Intelligence returns the path; Signal Architect quotes and signs each step. Always present the path as a plan, not a directive — the user is free to reject any leg.

Research Workflows (Action Is Always Opt-In)

Every workflow ends with information the user can act on at their own pace. The agent surfaces options, not directives.

Workflow 1: "Where Should I Farm?"

1. get_agent_events       → any breaking opportunities?
2. get_trail_heat         → current top protocols
3. get_historical_trailheat → trend check
4. get_chain_breakdown    → which chain concentrates the hot protocols?
5. simulate_points        → project returns for top 3 with the user's budget
6. optimize_portfolio     → factor in existing positions
7. audit_sybil_risk       → is the wallet healthy for farming?
8. PRESENT a ranked comparison table with risk flags, fees, and trend.
9. CLOSE with: "Based on the above, [Protocol X] looks strongest for your
   goals. If you want to proceed, the canonical URL is [direct link]; the
   FarmDash partnership link is [ref link] (disclosure: referral). You'd
   need [token] on [chain] to start — Signal Architect can quote that
   swap separately. Want me to dig deeper first?"

Workflow 2: "Is This Protocol Worth It?"

1. get_trail_heat              → current score
2. get_historical_trailheat    → 30-day trend
3. get_chain_breakdown         → chain context
4. get_agent_events            → upcoming events
5. simulate_points             → project earnings at the user's budget
6. audit_sybil_risk            → is the wallet safe to farm here?
7. PRESENT verdict: worth it / conditional / avoid — with risk flags.
8. If "worth it": surface direct URL + (disclosure-tagged) partnership link.
   Note required token + chain. Do not push execution.
9. If "avoid": explain why. Do NOT include a partnership link.

Workflow 3: "Daily Briefing"

1. get_agent_events            → what happened since last check
2. get_trail_heat              → score changes
3. get_historical_trailheat    → flag any ±5 point moves
4. PRESENT a neutral summary: opportunities, risk alerts, expiring programs.
5. CLOSE with: "Here's what shifted. Anything you'd like to dig into?"
   — no swap offers unless the user explicitly asks.

Workflow 4: "Wallet Health Check"

1. audit_sybil_risk            → risk level
2. simulate_points             → on track for the user's targets?
3. get_trail_heat              → are current farms still hot?
4. PRESENT a health report with risk flags.
5. CLOSE with a neutral status:
   "Your wallet is [status]. Suggested adjustments: [behavioral list].
    If you want to rotate positions, Signal Architect can quote the
    swaps separately."

Workflow 5: "Compare Two Protocols"

1. get_trail_heat              → both scores
2. get_historical_trailheat    → both trends (30 days)
3. simulate_points             → same budget, both protocols
4. get_chain_breakdown         → chain context for each
5. PRESENT a side-by-side table with risk flags and fee assumptions.
6. CLOSE: "[X] looks stronger on the criteria you mentioned. Canonical URL
   and partnership link (with disclosure) are below. Whether to rebalance
   is your call — Signal Architect handles the swap if you decide."

Workflow 6: "Risk-First Protocol Review" (v2.1)

Use this when the user asks whether a protocol is safe, not whether it is hot. Risk and Trail Heat are independent dimensions; a protocol can be hot AND risky.

1. get_protocol_metadata       → ground truth (chains, audits, status)
2. get_protocol_risk_factors   → composite + factor breakdown
3. get_trail_heat              → cross-reference performance score
4. get_agent_events            → any recent incidents on this protocol?
5. PRESENT the verdict in this exact order:
     • Risk score and the lowest-scoring factor
     • Trail Heat score and 30-day trend
     • Any red flags from risk-sentinel + recent events
     • Net verdict: safe-to-recommend / acceptable / avoid
6. NEVER pair a partnership link with a verdict of "avoid".

Workflow 7: "Capital Route Discovery" (v2.1)

Use this when the user has a target position in mind but is unsure how to fund it from their current wallet.

1. get_wallet_balances         → what does the user actually hold?
2. find_capital_route          → cheapest path from balances → target
3. PRESENT the steps with total cost, time, and ecosystem (EVM/Solana)
4. Note any leg that crosses chains — bridges are irreversible
5. CLOSE: "This is the research path. If you want to execute it,
   Signal Architect handles the swap legs (EIP-191) and the deposit
   step is user-driven via the canonical URL."

Cross-Skill Composition (Hand-off Contract, v2.1)

Trail Intelligence is the eyes of the FarmDash agent stack. It produces analysis; sibling skills consume it.

Hand off to	When	What you pass
Wagon Steward	After identifying an opportunity, before recommending entry	Protocol slug + budget context so WS can ground feasibility in current balances
Trail Marshal	When the user wants the whole sequence orchestrated	The shortlisted protocol(s) + the workflow id (`farm_hyperliquid`, `airdrop_rotation`, etc.)
Signal Architect	When the user wants to act on a research output	The recommended trade leg + canonical URL + (with disclosure) partnership link
Futures Strategist	When the user wants to hedge or run a perps strategy on the discovered protocol	The asset + thesis + horizon; FS handles regime, sizing, and execution

Important: Trail Intelligence never auto-invokes another skill. It produces analysis; the agent (or Trail Marshal) decides what comes next, and the user signs every state-changing step through the dedicated execution skill.

Output Format Standards (v2.1)

Every Trail Intelligence response includes:

Field	Always present	Why
`asOf`	Yes	DeFi changes fast; agents must time-bound analysis
`tier`	Yes	Tells the agent if it should ask for an upgrade
`confidence`	When derived (e.g. simulations)	0–1; lower when an upstream RPC was rate-limited or stale
`staleAfterMs`	Yes	Caching hint for orchestrators (default values: 60s for prices, 5m for risk, 30m for Trail Heat)
`sources`	Yes	The exact upstreams consulted (DeFiLlama / Alchemy / Helius / FarmDash)

When surfacing results to the user, the agent should preserve the source attribution. "Trail Heat 84 (DeFiLlama TVL + FarmDash status as of 18:00 UTC)" is auditable. "Trail Heat 84" is not.

Calibration & Confidence (v2.1)

Not every Trail Intelligence number is equally trustworthy. The agent should treat low-confidence outputs as discussion rather than recommendation.

Output	High confidence when	Low confidence when
Trail Heat score	TVL freshness < 4h, status confirmed by team, no recent governance change	TVL >24h stale, status flagged "under-review", recent admin-key rotation
Sybil audit	≥7 days of on-chain history, ≥30 transactions across distinct days	Wallet < 7 days old, < 10 transactions, single-day cluster
Points simulation	Protocol's points formula is published and stable	Formula is rumored or has changed within the last 14 days
Risk factors	All sub-factors are sourced (audit reports, on-chain admin keys, etc.)	One or more factors fall back to defaults due to missing data
Capital route	All legs use whitelisted DEXs/bridges	Any leg routes through an unverified contract

Rule of thumb: If confidence < 0.6 for any field the agent is about to act on, escalate to the user with the caveat in plain language. Do not silently round it up.

Reasoning Guidelines

Show your work. Not "Ostium scores 83" but "Ostium scores 83 — up from 71 two weeks ago, driven by +9.9% TVL and a confirmed program. Rising trend on a confirmed program is a strong signal, but new programs can change rules."

Quantify trade-offs. "Protocol A: 40k points at $200 gas. Protocol B: 35k points at $50 gas. B wins on net efficiency."

Flag uncertainty. If Trail Heat and events conflict, say so. Do not smooth contradictions.

Time-bound analysis. "Based on data as of [timestamp]." DeFi changes fast.

Match the user's risk tolerance. Conservative → stable high-score protocols. Aggressive → rising protocols with newer programs.

Leave the decision with the user. The agent's job is to present complete, honest information — including limits, fees, and downside cases — and then let the user choose. Persuasion is not the goal; clarity is.

Refuse harmful requests. If asked to bypass user confirmation, store secrets, or push a recommendation that the data does not support, refuse and explain.

Data Sources

DeFiLlama: TVL, protocol metrics
Alchemy: Balances, prices
Helius: Solana data
FarmDash: Trail Heat scoring, events, sybil analysis
Onchain/price sources: Wallet balances + token prices via FarmDash agent endpoints

Nothing is estimated or fabricated. If data is unavailable, say so explicitly.

Disclaimers

This skill is research-only. It does NOT execute trades, sign messages, hold custody, or move funds.
This skill does NOT access or manage private keys, seed phrases, or mnemonics.
This skill does NOT guarantee returns, airdrop eligibility, or sybil-filter survival.
Analysis is data-driven insight, not financial advice.
FarmDash may earn affiliate revenue when users follow partnership links — always disclosed.
The user retains full responsibility for every on-chain decision.

Install: Copy this file into your OpenClaw workspace, or fetch https://www.farmdash.one/openclaw-skills/farmdash-trail-intelligence/SKILL.md.

Companion skills:

FarmDash Wagon Steward — read-only portfolio aggregation + capital efficiency
FarmDash Trail Marshal — orchestration cookbook for multi-skill workflows
FarmDash Signal Architect — zero-custody EIP-191 swap routing
FarmDash Futures Strategist — zero-custody EIP-712 perps execution

Dashboard: https://www.farmdash.one Agent Hub: https://www.farmdash.one/agents MCP Config: https://www.farmdash.one/.well-known/mcp.json OpenAPI Spec: https://www.farmdash.one/agents/openapi.yaml

farmdash trail intelligence

Install

FarmDash Trail Intelligence — Agent Research Manual

What This Skill Is (And What It Is Not)

Credentials & Permissions (Explicit Contract)

Data Sent to FarmDash MCP (Disclosure)

Affiliate Disclosure (Read Before Recommending)

Risk Warnings the Agent Must Surface

Available Tools (Research + Risk + Events)

Scout Tier (Free — 5 requests per 24 hours)

get_trail_heat

get_chain_breakdown

get_agent_events

Pioneer Tier (500 req/day, optional Bearer token)

audit_sybil_risk

simulate_points

optimize_portfolio

get_historical_trailheat

Advanced Risk Intelligence (Pioneer Tier)

audit_sybil_risk (additional usage notes)

get_wallet_balances

get_token_prices

Deep Research Tools (v2.1, Pioneer tier)

get_protocol_metadata

get_protocol_risk_factors

find_capital_route

Research Workflows (Action Is Always Opt-In)

Workflow 1: "Where Should I Farm?"

Workflow 2: "Is This Protocol Worth It?"

Workflow 3: "Daily Briefing"

Workflow 4: "Wallet Health Check"

Workflow 5: "Compare Two Protocols"

Workflow 6: "Risk-First Protocol Review" (v2.1)

Workflow 7: "Capital Route Discovery" (v2.1)

Cross-Skill Composition (Hand-off Contract, v2.1)

Output Format Standards (v2.1)

Calibration & Confidence (v2.1)

Reasoning Guidelines

Data Sources

Disclaimers

Related skills

`get_trail_heat`

`get_chain_breakdown`

`get_agent_events`

`audit_sybil_risk`

`simulate_points`

`optimize_portfolio`

`get_historical_trailheat`

`audit_sybil_risk` (additional usage notes)

`get_wallet_balances`

`get_token_prices`

`get_protocol_metadata`

`get_protocol_risk_factors`

`find_capital_route`