#farmdash #wagon-steward #capital-efficiency #defi

FarmDash Wagon Steward

Use when watching a DeFi portfolio: balances, position health, idle capital, drift. Read-only; never rebalances or moves funds without your signature.

FarmDash@parmasanandgarlic

Install

openclaw skills install @parmasanandgarlic/farmdash-wagon-steward

FarmDash Wagon Steward

Use this skill when asking "what is happening in my wallet": balances, idle capital, allocation drift, and rebalance proposals — read-only, address alone.

[!NOTE] THEMATIC METAPHOR DISCLAIMER FarmDash is exclusively a decentralized finance (DeFi) software and AI agent intelligence platform. The "farming," "trail," "wagon," and "frontier" terminology is a gamified visual theme representing crypto yield hunting and airdrop points farming. It does not relate to physical agriculture or agrifood industries.

[!IMPORTANT] ZERO-CUSTODY CRITICAL BOUNDARY This skill is strictly a read-only portfolio aggregation surface and does NOT execute any state-changing swaps, deposits, or perps, nor does it request, hold, or transmit private keys, seed phrases, or mnemonics.

Security Posture. Strict read-only. Every tool is a GET endpoint that returns analysis derived from public on-chain data. No environment variables are required for the free tier. The optional Bearer token raises rate limits only and is sent solely as an Authorization header to https://www.farmdash.one. Wagon Steward cannot modify wallet state, approve allowances, or initiate any on-chain action. Recommendations are research output; the user's separately-installed execution skill is responsible for any on-chain action under its own ClawScan-reviewed contract.

What this skill does

Wagon Steward is the read-only mirror of the user's farming state. Where Trail Intelligence answers "what is happening in DeFi," Wagon Steward answers "what is happening in your wallet."

It produces analysis, not transactions. Every output is a research snapshot the user reviews and can act on, at their own pace, through whichever execution skill they have separately installed.

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

Permissions

Wagon Steward requires only the user's public wallet address (the same value visible on any block explorer) and, for Pioneer-tier rate limits, an optional Bearer token. It does not request, accept, or transmit any other credentials. It defines no write capabilities. Recommendations are returned as JSON for the agent to surface to the user; if the user chooses to act, the action is performed by their own separately-installed execution skill under that skill's own ClawScan-reviewed contract.

Data sent to FarmDash MCP

Field	Sent?	Notes
Public wallet address (`0x…`)	Yes, every call	Public information — same as a block-explorer query
Chain IDs to inspect	Yes, optional	Public values
Bearer token, if provided	Yes, Authorization header	Tier identification only

No other fields are collected by skill logic.

Self-hosting option: Wagon Steward endpoints are documented in the FarmDash unified OpenAPI spec at https://www.farmdash.one/agents/openapi.yaml.

Tier Model

Tier	Cost	Limits	Capabilities
Scout	Free	5 req / 24h	Public FarmDash discovery only; wallet portfolio endpoints require Pioneer because they consume portfolio API credits
Pioneer	$39.99/mo	1,500 req / day	Wallet balances, portfolio summary, idle capital, token prices, position/performance health
Syndicate	$199/mo	50k req / day	Higher-rate portfolio loops, webhooks through Signal Architect, unrestricted CORS, and advanced session/control tooling for teams and serious agents

Available Tools

Current MCP Tool Map (v0.6)

The current FarmDash MCP server exposes these Wagon-relevant tools directly:

get_wallet_balances
get_portfolio_summary
get_position_health
get_idle_capital
get_token_prices
get_agent_performance

This SKILL.md also defines useful agent procedures named get_capital_efficiency, get_rebalance_plan, and watch_wagon. Preserve those names in reasoning and UI copy, but do not call them as MCP tools unless the runtime explicitly exposes them. Today they are derived procedures:

get_capital_efficiency = derive from get_portfolio_summary, get_idle_capital, concentration, priced tokens, and stale-position signals.
get_rebalance_plan = pair Wagon state with Signal Architect optimize_portfolio and fresh swap quotes.
watch_wagon = implement as an agent loop using repeated reads plus Signal Architect webhooks when the user has Syndicate access.

If a user asks for one of the derived procedures, execute the underlying current tools and label the result as "derived by the agent from FarmDash wallet data."

`get_portfolio_summary` (Pioneer)

Top-level snapshot. Use this first — it grounds every other recommendation.

Inputs: current MCP uses address (required EVM address) and optional chains (comma-separated Alchemy network IDs such as eth-mainnet,base-mainnet). Legacy examples may say walletAddress; map that to address before calling.

Current returns: address, queried Alchemy networks, portfolio.totalValueUsd, nativeBalanceUsd, erc20Count, chainsActive, chainDistribution, up to ten topTokens, and timestamp.

It does not currently identify protocol positions, liabilities, LP composition, locked/vesting balances, lending debt, perp exposure, cost basis, or cross-chain balances outside the requested/default network set. Treat totalValueUsd as a partial observable-wallet estimate, not net asset value.

Agent posture: Start a wagon-aware conversation with this tool when the user has Pioneer access. If the call returns 402, explain that wallet portfolio reads require Pioneer and continue with non-wallet public research instead of fabricating balances. Do not narrate the entire response; surface the totals plus any flags Wagon Steward returns or that you derive (over-concentration, large idle capital, stale positions).

`get_position_health` (compatibility alias; Pioneer)

Despite its legacy name, the current MCP mapping calls /api/v1/agent/performance. It returns FarmDash activity metrics (swap count, volume, fees, average size, protocol diversity, active days), Trail Cred/ranking data, daily activity, and a timestamp.

It does not return farming positions, APR, unrealized or realized P&L, impermanent loss, collateral health, liquidation distance, or rebalance hints. Never use it to claim a DeFi position is healthy. For a real position review, obtain protocol-native position/debt data and authoritative prices; if those tools are unavailable, say position health is unavailable.

`get_idle_capital` (Pioneer)

Surfaces stablecoins or unstaked native assets sitting unused.

Inputs: current MCP uses address. minThresholdUSD is a derived agent display filter; apply it after the response if needed.

Current returns: idleCapitalUsd, idleAssets, count, networks, and timestamp. The filter treats configured stablecoin symbols and native assets as potentially idle. It does not know reserves, collateral needs, pending transactions, tax obligations, user intent, or a top yield destination, and it does not calculate opportunity cost.

Agent posture: Call it "candidate liquid reserves," not idle capital, until the user confirms required gas, collateral, withdrawal, tax, and emergency reserves. Analysis may compare options; no automatic deployment intent.

`get_capital_efficiency` (derived procedure, not an MCP tool)

A client-side 0–100 heuristic for discussion. The current server does not return this score. It is not risk-adjusted return, Sharpe ratio, expected utility, or a mandate to deploy reserves.

Formula (transparent):

text

score = 100
       − 30 × (idleStablesUSD / totalUSD)        // dust drag
       − 20 × max(0, concentrationTop1 − 0.40)   // over-concentration
       − 15 × max(0, dryProtocolsCount × 0.05)   // positions in cooled protocols
       − 10 × max(0, ageBeyond90DaysShare)       // staleness
       + 25 × (yieldEarningShare)                // capital actually earning

Returns:

text

{
  "score": 72,
  "breakdown": {
    "earning": 0.62,
    "idle": 0.15,
    "suboptimal": 0.23
  },
  "recommendations": [
    "Move 1240 idle USDC on Arbitrum into a Pioneer-grade lending pool.",
    "Top-1 concentration is 49.7% — diversify before adding.",
    "Uniswap V3 LP is trending below median; consider reducing 20%."
  ]
}

Agent posture: Show every input and exclude unpriced/missing assets from the calculation rather than assigning zero. Do not lead with this score when coverage is incomplete, and never let it override user reserve requirements or protocol risk.

`get_rebalance_plan` (derived procedure, not an MCP tool)

Pair current wallet data with optimize_portfolio and fresh quotes to draft a proposal. The illustrative schema below is not a current Wagon endpoint response.

Inputs:

text

{
  "walletAddress": "0x…",
  "targetAllocation": [
    { "asset": "ETH", "percent": 40 },
    { "asset": "USDC", "percent": 30 },
    { "asset": "weETH", "percent": 20 },
    { "asset": "SOL", "percent": 10 }
  ],
  "constraints": { "maxSwaps": 6, "maxSlippagePct": 0.5 }
}

Returns:

text

{
  "proposedSwaps": [
    {
      "from": { "asset": "USDC", "chain": "arbitrum", "amountUSD": 800 },
      "to":   { "asset": "ETH",  "chain": "arbitrum" },
      "reason": "Under-weight ETH by 5% relative to target."
    },
    ...
  ],
  "expectedFeeUSD": 18.40,
  "expectedSlippageUSD": 4.20,
  "netDriftClosedPct": 9.4
}

Agent posture: Present the plan as a discussion, not a directive. Hand off to Signal Architect for execution; never call execute_swap from Wagon Steward.

`watch_wagon` (derived procedure, not an MCP tool)

Implement only in a client/orchestrator that performs repeated reads and owns its webhook lifecycle. The current MCP registry does not expose watch_wagon, and Wagon cannot claim a server subscription exists.

Inputs:

text

{
  "walletAddress": "0x…",
  "alertWebhook": "https://your-agent.example.com/webhook",
  "thresholds": {
    "concentrationTop1": 0.50,
    "idleCapitalUSD": 2000,
    "drawdown7dPct": 5
  }
}

The example input below is a design pattern, not a promise of a watchId, 15-minute cron, or wagon.event.* server events.

Cross-Skill Composition

Wagon Steward is designed to compose with the other FarmDash skills. The standard composition pattern:

text

Trail Intelligence  →  Wagon Steward  →  Signal Architect / Futures Strategist
   "what's hot"        "what I have"       "make it happen"

Companion skill	When to call it	Hand-off pattern
Trail Intelligence	Before recommending any new entry — confirm the protocol is still hot	TI returns Trail Heat + risk → WS confirms feasibility against current balances
Trail Marshal	When the user wants the whole sequence orchestrated	TM calls WS first to ground the plan in real balances, then calls SA/FS to execute
Signal Architect	When the rebalance plan needs to actually move tokens	WS produces the plan, SA quotes + executes each swap with user confirmation
Futures Strategist	When the user wants to hedge observable spot exposure	WS passes partial spot context; Hedge Warden/FS plan a candidate hedge and post-fill reconciliation measures residual delta

Important: Wagon Steward never auto-calls another skill. It produces analysis; the user (or an orchestrator like Trail Marshal) decides what to do with it.

Report-Back Template — State / Risks / Invalidation / Costs

Report: address + timestamp/asOf + queried networks; totalValueUsd as partial observable-wallet estimate; topConcentrationPct vs 40% gate; idleCapitalUsd as candidate reserves; priced/unpriced counts; overConcentrated / idleDrag / staleData / routeNeeded flags; invalidation (missing chains, unpriced assets, unavailable debt/perp data blocks net-exposure claims); costs; nextBestAction (hold / monitor / research / quote_swap / reduce / rebalance) + handoff (none / trail_intelligence / trail_marshal / signal_architect / futures_strategist).

Output and Coverage Standards

Every Wagon Steward response includes:

Field	Always present	Why
`timestamp` / normalized `asOf`	When returned	Time-bound the snapshot; never invent source time
`address`	Yes for wallet calls	Prevent context confusion
queried networks	Yes	Make coverage boundaries visible
priced/unpriced counts and values	Derive when possible	Missing/unpriced assets are unknown, not zero
confidence	Only when methodology is stated	Never present an arbitrary 0–1 value as API output
freshness limit	Client policy	State the policy separately from source data

Risk Warnings the Agent Must Surface

For any portfolio recommendation the agent should restate that:

Balances are read at a moment in time; mempool activity may shift them within minutes
Cross-chain holdings may include unverified tokens — Wagon Steward only scores tokens it can price
Rebalance plans assume current liquidity; real execution may have higher slippage on volatile pairs

Rebalance Cost-Drag Checklist

For every get_rebalance_plan discussion report: pre/post-trade weights using conservative executable prices; expectedFeeUSD + expectedSlippageUSD; maxSwaps and maxSlippagePct constraints; netDriftClosedPct; rebalance bands and minimum trade sizes to prevent churn; tax, impermanent loss, and gas notes; and coverage gaps (unpriced assets, missing chains, debt, LP look-through, locked balances). Hand off quotes and signatures to Signal Architect.

Capital efficiency is a heuristic, not financial advice
The user is solely responsible for the final decision

Autonomous Portfolio Intelligence Upgrade (v0.6)

Every Wagon-aware agent loop should maintain a portfolio state ledger. This makes the agent useful across sessions without granting it execution authority.

json

{
  "walletAddress": "0x...",
  "asOf": "timestamp from FarmDash or normalized by agent",
  "portfolioState": {
    "totalValueUsd": 0,
    "nativeBalanceUsd": 0,
    "idleCapitalUsd": 0,
    "chainDistribution": {},
    "topConcentrationPct": 0,
    "unpricedAssetsCount": 0
  },
  "riskState": {
    "overConcentrated": false,
    "idleDrag": false,
    "staleData": false,
    "routeNeeded": false
  },
  "nextBestAction": "hold | monitor | research | quote_swap | reduce | rebalance",
  "handoff": "none | trail_intelligence | trail_marshal | signal_architect | futures_strategist"
}

Decision rules:

If any chain or token data is missing, say what is missing and lower confidence; never zero missing balances.
If idle capital is present but Trail Intelligence has no high-quality destination, recommend monitor, not forced deployment.

Deploy-or-Hold Playbook for Candidate Reserves

After get_portfolio_summary + get_idle_capital: confirm gas, collateral, withdrawal, tax, and emergency reserves; call idle funds candidate liquid reserves until confirmed; check Trail Intelligence Trail Heat + risk before any entry; apply the 40% top-concentration gate before adding to the same asset or chain; if no high-quality destination, return monitor with nextBestAction hold / monitor / research and handoff none / trail_intelligence.

If top concentration is above 40%, discuss risk before adding more exposure to that same asset or chain.
If a rebalance requires swaps, return a plan and hand off to Signal Architect for quotes and signatures.
If spot exposure needs a hedge, pass the asset, notional size, and risk concern to Futures Strategist; do not size perps from this skill alone.

Portfolio Risk Contract

Before recommending new risk, create a coverage table for observable assets, unpriced assets, missing chains, debt/liabilities, LP look-through, locked balances, and off-wallet/perp exposure. If any material category is unavailable, label portfolio NAV and concentration as partial.

Apply limits hierarchically:

preserve explicit gas, collateral, withdrawal, tax, and emergency reserves;
measure concentration by asset, stablecoin issuer, protocol, chain, bridge, venue, and correlated factor—not token symbol alone;
look through LP, vault, LST/LRT, and receipt tokens when authoritative composition exists;
calculate pre/post-trade weights using conservative executable prices and include fees/slippage;
use rebalance bands and minimum trade sizes to prevent churn;
never call a portfolio delta-neutral until actual spot and hedge fills are reconciled and residual delta is measured.

Unknown balances or liabilities block claims about net exposure. A wallet-balance snapshot alone cannot establish solvency, P&L, or complete portfolio risk.

Watch Threshold Triage (Derived-Loop Only)

In watch_wagon derived loops using repeated reads, triage concentrationTop1 at 0.50, idleCapitalUSD at 2000, and drawdown7dPct at 5 as review triggers, not execution triggers; on trigger, re-run get_portfolio_summary, derive coverage-aware capital efficiency only if inputs are complete, exclude unpriced/missing assets rather than assigning zero, lower confidence on missing data, and route hedge questions to Futures Strategist with asset, notional, and risk concern.

Recommended Workflows

Workflow 1: "Wagon Briefing"

text

1. get_portfolio_summary           → totals + top positions
2. DERIVE coverage-aware capital efficiency only if inputs are complete
3. get_idle_capital                → candidate liquid reserves; confirm reserve requirements
4. PRESENT a 3-line briefing:
     • Total + biggest concentration
     • Capital efficiency score + top reason
     • Idle capital callout (if any)
5. CLOSE: "Anything you'd like to dig into?"

Workflow 2: "Healthcheck Before Adding"

text

1. get_portfolio_summary
2. get_position_health → activity metrics only; obtain protocol-native position health separately
3. PRESENT observable concentration + explicit coverage gaps
4. NOTE if the new entry would push any concentration above 40%
5. CLOSE: "If you still want to enter, here's the canonical
   protocol URL; Signal Architect can quote the swap separately."

Workflow 3: "Quarterly Rebalance"

text

1. get_portfolio_summary
2. Obtain protocol-native position/debt health; do not use the compatibility activity endpoint as position health
3. (optional) Trail Intelligence: get_trail_heat for current positions
4. get_rebalance_plan with the user's target allocation
5. PRESENT the plan as a discussion: which swaps, why, fees, drift closed
6. CLOSE: "Want Signal Architect to quote these swaps so you can sign them?"

Workflow 4: "Idle Capital Deploy"

text

1. get_idle_capital → candidate reserves, pending user reserve confirmation
2. Supply Master: compare_yields across eligible pools
3. PRESENT comparison: base/reward APY, stability, TVL, protocol/asset risk, all costs, exit constraints, and missing evidence
4. CLOSE: "I can hand off to Signal Architect to position you. Want to proceed?"

Reasoning Guidelines

Show your work. Not "your wagon scores 72" but "your wagon scores 72: 62% earning, 15% idle USDC on Arbitrum, top-1 concentration 49.7% (target ≤40%)."

Quantify trade-offs. "Moving idle USDC into Aave at 12% APR earns ~$148/year. The swap fee + bridge cost would be ~$8 — break-even in 20 days."

Flag uncertainty. If a chain RPC fails or a token can't be priced, say so explicitly. Do not zero out missing data.

Time-bound analysis. "Snapshot as of [timestamp]. Trail Heat scores can shift by tomorrow."

Never editorialize the score. The capital-efficiency formula is published. Quote it neutrally; let the user judge.

Refuse harmful requests. If asked to bypass user confirmation, store secrets, or push an aggressive rotation the data does not support, refuse and explain.

Disclaimers

This skill is read-only. It does NOT execute trades, sign messages, hold custody, or move funds.
This skill does NOT access private keys, seed phrases, or mnemonics.
Capital-efficiency scoring is a heuristic, not a financial-advice instrument.
Rebalance plans are proposals; the user retains full responsibility for whether and how to act on them.
Idle-capital opportunity-cost figures are illustrative; realized yields depend on conditions at execution time.

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

Companion skills:

FarmDash Trail Intelligence — DeFi research and Trail Heat scoring
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

Why FarmDash: Unlike portfolio trackers that demand exchange keys, Wagon Steward mirrors your farming state from a public address alone — nothing to connect, nothing to custody, nothing to leak.

FarmDash: Multichain DeFi portfolio intelligence Agent Hub: FarmDash portfolio analytics for autonomous agents OpenAPI Spec: FarmDash API Schema MCP Config: FarmDash MCP Server

Optional Setup Check (Consent-Gated & Non-State-Changing)

[!IMPORTANT] PRIVACY & OPTIONAL TELEMETRY WARNING Running this optional setup check communicates with a third-party server (https://www.farmdash.one). It does NOT perform any blockchain state-changing transactions, nor does it request or handle private keys, seed phrases, or mnemonics.

Data Disclosed: If executed, this command sends your public agent/wallet address and the skill identifier to FarmDash's analytics server to identify your active subscription tier (Scout/Pioneer/Syndicate) and returns your configuration capabilities.

Gated Consent Protocol:

This check is completely optional. Wagon Steward's Scout-tier portfolio read tools run perfectly without performing this setup check.

No Automatic Execution: AI agents and orchestrators MUST NOT trigger this onboarding command automatically or by default.

Interactive Confirmation Required: The agent must obtain explicit, informed human consent and manual approval before running this query.

Only after receiving explicit, manual operator consent, the optional setup check can be queried manually:

bash

curl -X POST https://www.farmdash.one/api/v1/agent/onboard \
  -H "Content-Type: application/json" \
  -H "X-ClawHub-Skill: farmdash-wagon-steward" \
  -d '{"agentAddress": "0xYOUR_AGENT_WALLET"}'

What this optional query does:

Retrieves Configuration: Returns your current tier access (Scout/Pioneer/Syndicate) and active tool definitions.
Validates API Tier Limits: Connects your ClawHub session to the corresponding subscription limits for portfolio reading.
Analytics Registration: Registers your public agent wallet address in the FarmDash service funnel solely for capacity planning and active session telemetry.

Next steps after manual check:

GET /api/v1/agent/onboard — Full capability map and setup guide.
Upgrade to Pioneer ($39.99/mo) or Syndicate ($199/mo) for execution access.
Browse the OpenAPI spec at https://www.farmdash.one/agents/openapi.yaml. Send X-ClawHub-Skill: farmdash-wagon-steward on every FarmDash API request — not just the optional check above — so anonymous Scout usage is attributed to this skill. Analytics-only and optional; requests without the header still work. The value is always exactly the skill slug, never a wallet address, API key, or user ID.

FarmDash Wagon Steward

Install

FarmDash Wagon Steward

What this skill does

Permissions

Data sent to FarmDash MCP

Tier Model

Available Tools

Current MCP Tool Map (v0.6)

get_portfolio_summary (Pioneer)

get_position_health (compatibility alias; Pioneer)

get_idle_capital (Pioneer)

get_capital_efficiency (derived procedure, not an MCP tool)

get_rebalance_plan (derived procedure, not an MCP tool)

watch_wagon (derived procedure, not an MCP tool)

Cross-Skill Composition

Report-Back Template — State / Risks / Invalidation / Costs

Output and Coverage Standards

Risk Warnings the Agent Must Surface

Rebalance Cost-Drag Checklist

Autonomous Portfolio Intelligence Upgrade (v0.6)

Deploy-or-Hold Playbook for Candidate Reserves

Portfolio Risk Contract

Watch Threshold Triage (Derived-Loop Only)

Recommended Workflows

Workflow 1: "Wagon Briefing"

Workflow 2: "Healthcheck Before Adding"

Workflow 3: "Quarterly Rebalance"

Workflow 4: "Idle Capital Deploy"

Reasoning Guidelines

Disclaimers

Optional Setup Check (Consent-Gated & Non-State-Changing)

What this optional query does:

Next steps after manual check:

Official FarmDash Links

Related skills

`get_portfolio_summary` (Pioneer)

`get_position_health` (compatibility alias; Pioneer)

`get_idle_capital` (Pioneer)

`get_capital_efficiency` (derived procedure, not an MCP tool)

`get_rebalance_plan` (derived procedure, not an MCP tool)

`watch_wagon` (derived procedure, not an MCP tool)