Ready to supercharge your AI agents?

Install any skill in seconds. Free to start, no credit card required.

Lista Lending Skill (POC) · leoyeai | DecimalAI

leoyeai/lista-lending

SkillScore36Maintained

View and operate Lista Lending vaults/markets. Use when user asks about LENDING-ONLY positions or wants to deposit/withdraw/borrow/repay. For report-style overview, use lista instead.

financeby leoyeai

Links this skill to your agents and keeps it current with the author’s updates.

Automatic

Finance & Accounting pack

1 · One-time — install the CLI (docs ↗)

pip install decimalai

2 · Get the skill — writes .claude/skills/leoyeai-lista-lending/SKILL.md

decimalai skills pull leoyeai-lista-lending --out .claude/skills/

No CLI? Get the same file — curl · raw file

curl -fsSL --create-dirs -o .claude/skills/leoyeai-lista-lending/SKILL.md https://app.decimal.ai/s/leoyeai-lista-lending/SKILL.md

Raw file: /s/leoyeai-lista-lending/SKILL.md ↗ · machine-readable summary at /s/leoyeai-lista-lending.json ↗

In productionGlobal

how this skill behaves on real traffic · last 30 days

No production data for this skill yet — usage numbers appear here once agents start running it.

Anonymized aggregate across every org using this skill — proof from real production use, not a demo.

Does this skill help? · Benchmark: with vs. withoutsign in to run this on your agents

25 test cases from the publisher’s eval.yaml · each run twice on gemini-3.6-flash — with the skill injected, then without · last run 50d ago

—+280% tokens0% turns36% of whole cases pass with skill (9/25)

Test case	Without → With	Effect	Δ tokens	Δ turns
case-05	✗→✓	▲ Improved	143%	0%
case-08	✗→✓	▲ Improved	279%	0%
case-11	✗→✓	▲ Improved	890%	0%
case-09	✓→✗	▼ Worse	658%	0%
case-04	✓→✓	＝ Same ✓	346%	0%

Showing 5 of 25 cases

SkillSafety

✓ 3 of 3 passed

Code scan15 of 15 checks clean

0 critical · 0 warning · 0 info — scanner v4 — 8/2/2026

live_secretremote_code_executionreverse_shelldata_exfiltrationssrf_metadataagent_snoopinganti_refusaldestructive_commandsobfuscated_unicodeunicode_homoglyphopaque_blobsuspicious_urltool_overreachtrigger_abuseprompt_injection_phrasing

IntentSafe

LLM judge · does the skill do what it claims

ContentClean

LLM judge · policy & harmful-content review

Source

Imported

LeoYeAI / openclaw-master-skills

skills/lista-lending/SKILL.md

View on GitHub ↗

License MITImported 8/2/2026Version v1

Skill contents

SKILL.md

Lista Lending Skill (POC)

> Status: Pre-release (internal only, not in public registry) > Agent quick check: node dist/cli/cli.bundle.mjs version > Agent rebuild: npm install && npm run build

Execute Lista Lending vault and market operations through @lista-dao/moolah-lending-sdk, with transaction sending delegated to lista-wallet-connect.

Agent Execution Policy

Execute CLI/setup commands directly as the agent; do not ask the user to run shell commands.
If dependencies/build are missing or outdated, run setup/build automatically.
Before any signing or on-chain write action, explain the action and get user consent.
Do not output raw JSON/internal payloads to users by default; convert results into human-readable summaries/tables.
Use command snippets in this document as agent-side execution references only.

When to Use This Skill

This skill handles LENDING-SPECIFIC viewing and operations.

✅ Use when user wants to:

1. View LENDING positions only:

"Check my lending positions"
"Show my Lista vaults"
"What's in my lending markets"
"My collateral in lending vaults"

2. Execute lending operations:

"Deposit 100 USDT to Lista vault"
"Withdraw from my Lista vault"
"Borrow against my collateral"
"Repay my lending debt"
"Find best lending yield"

❌ Do NOT use for:

All-product/report-style overview → Use lista
"Check ALL my Lista positions" (report view) → Use lista
"My total Lista collateral/debt" (summary-style answer) → Use lista

Key difference:

lista: VIEW report-style lending analytics and summaries
lista-lending: VIEW/OPERATE lending only (vaults + markets)
lista-wallet-connect: Wallet operations (connect/sign/call)

Routing Priority (Required)

Apply these routing rules before running commands:

Wallet/session/signing issues (pair, auth, sign, topic, session) → lista-wallet-connect
Lending write operations (deposit, withdraw, supply, borrow, repay, market-withdraw) → lista-lending
Lending actionable reads (holdings, vaults, markets for selection/filtering/follow-up execution) → lista-lending
Report/narrative analysis (position report, risk check, daily digest, high-level market/vault summary) → lista

Ambiguous intent handling:

check my positions: if user asks for operation follow-up or lending-only detail, use lista-lending; otherwise use lista.
market list / vault list: if user needs execution target selection, use lista-lending; if user only wants macro overview, use lista.

Scope

Supported capabilities:

Vault: vaults, select --vault, deposit, withdraw
Market: markets, select --market, supply, borrow, repay, market-withdraw
Portfolio: holdings
Runtime/config: config, version

Not in scope:

CDP operations
Non-EVM chains

Temporary Limitations

Current unsupported market types for trade actions (select/supply/borrow/repay/market-withdraw):

SmartLending markets (zone === 3)
Fixed-term markets (termType === 1)

Behavior:

markets command filters them out from list output.
select --market rejects them with unsupported_market_type.
holdings still returns them (for full portfolio visibility) and marks:
isSmartLending
isFixedTerm
isActionable

Project Structure

text
skills/lista-lending/
├── SKILL.md
├── package.json
├── tsconfig.json
└── src/
    ├── cli/                # args/help/meta/run
    ├── api/                # vault/market/user read-side queries
    ├── sdk/                # SDK client + market runtime prefetch
    ├── executor/           # tx execution + receipt polling
    ├── commands/           # command handlers
    │   ├── shared/         # context/errors/output/tx helpers
    │   ├── select/
    │   ├── borrow/
    │   └── repay/
    ├── presenters/
    ├── utils/
    ├── config.ts
    ├── context.ts
    ├── api.ts              # API facade export
    ├── sdk.ts              # SDK facade export
    ├── executor.ts         # executor facade export
    └── cli.ts              # CLI entrypoint

Notes:

CLI parsing/routing lives in src/cli/* and keeps src/cli.ts as the single entrypoint.
Read-side querying is split by domain in src/api/* (vault, market, user).
SDK integration is centralized in src/sdk/client.ts; market runtime prefetch is in src/sdk/market-runtime.ts.
borrow and repay are split into dedicated simulate and execute modules to keep command files small.

Prerequisites

Wallet is paired via lista-wallet-connect skill.
You have a valid wallet-topic and wallet-address.
Node.js version is >= 18.0.0 (recommended >= 20).
lista-wallet-connect is built (skills/lista-wallet-connect/dist/cli/cli.bundle.mjs exists).
lista-lending is built (skills/lista-lending/dist/cli/cli.bundle.mjs exists).

Setup (Agent-Run)

Use prebuilt dist/ by default for faster startup.

bash
cd skills/lista-lending
# ensure Node.js >= 18 (recommended >= 20)
node -v
node dist/cli/cli.bundle.mjs version

Only run install/build when dist/ is missing or version check fails:

bash
cd skills/lista-lending
npm install
npm run build
node dist/cli/cli.bundle.mjs version

Fallback for local debugging (non-bundled output):

bash
cd skills/lista-lending
npm run build
node dist/cli.js version

If skills/lista-wallet-connect/dist/cli/cli.bundle.mjs is missing, rebuild wallet-connect as well:

bash
cd skills/lista-wallet-connect
npm install
npm run build

Agent execution path:

bash
cd skills/lista-lending
node dist/cli/cli.bundle.mjs <command> [options]

Output Contract

stdout: machine-readable JSON (result payload)
stderr: errors only
Optional debug mode: add --debug-log-file <path> to append structured stdout/stderr logs (jsonl).

Agent Display Guidelines

For user-facing answers, keep CLI JSON as internal source of truth, then render human-readable tables/summaries. Return raw JSON only when the user explicitly asks for raw output.

vaults recommended columns:

Vault, Asset, TVL (USD), APY, Curator

markets recommended columns:

Collateral, Loan, LLTV, Borrow Rate, Liquidity (USD)

holdings recommended structure:

Summary block first: total vault count, total market count, actionable market count, unsupported market count.
Vault table columns:
Vault, Chain, Deposited, Deposited USD, APY, Wallet Balance
Actionable market table columns:
Market, Chain, Collateral USD, Borrowed USD, LTV, Health
Unsupported market table columns (non-technical by default):
Market, Chain, Reason, Collateral USD, Borrowed USD

Default rule:

Use different table layouts for different commands (vaults, markets, holdings); do not reuse holdings layout for vault/market list pages.
Keep CLI/runtime JSON and debug logs in raw format (for example eip155:56, eip155:1) for stable machine parsing and debugging.
For user-facing text/tables, map chain IDs to names (eip155:1 -> Ethereum, eip155:56 -> BSC).
Do not show zone/termType in user-facing table unless user explicitly asks for technical details.
Do not paste full raw payloads (JSON/RPC internals) in normal user-facing replies.
Derive Reason from flags:
isSmartLending => SmartLending
isFixedTerm => Fixed-term

Health definition (for market positions):

Formula: health = LLTV / LTV when LTV > 0; otherwise health = 100.
Display label suggestion:
Healthy: health >= 1.2
Warning: 1.0 <= health < 1.2
Risk: health < 1.0
Action hint: if label is Warning or Risk, suggest repay/reduce borrow.

Supported Chains

eip155:56 (BSC, default)
eip155:1 (Ethereum)

Command Index

version - show skill version and compatibility hints
config - read/update RPC settings
vaults - list vaults with filters
markets - list markets with filters (SmartLending/fixed-term filtered)
holdings - query user positions across vault + market
select - set active vault or market context
deposit - deposit to selected/explicit vault
withdraw - withdraw from selected/explicit vault
supply - supply collateral to selected/explicit market
borrow - simulate or execute borrow
repay - simulate or execute repay
market-withdraw - withdraw supplied collateral from selected/explicit market

Global/Shared Options

--chain <eip155:56|eip155:1>
--wallet-topic <topic>
--wallet-address <0x...>

Command Details

All command snippets below are for agent execution; do not instruct the user to type them manually.

1) `version`

Purpose: Print skill version and dependency constraints.

bash
node dist/cli/cli.bundle.mjs version

2) `config`

Purpose: Manage RPC override.

Examples:

bash
# show config
node dist/cli/cli.bundle.mjs config --show

# set rpc override
node dist/cli/cli.bundle.mjs config --set-rpc --chain eip155:56 --url https://bsc-mainnet.nodereal.io/v1/<key>

# clear rpc override
node dist/cli/cli.bundle.mjs config --clear-rpc --chain eip155:56

Notes:

Config file: ~/.agent-wallet/lending-config.json
Chain fallback RPCs are still used if primary RPC fails.

3) `vaults`

Purpose: Discover vaults from SDK list API.

Common options:

--page, --page-size, --sort, --order
--zone, --keyword
--assets <a,b>, --curators <a,b>

Examples:

bash
node dist/cli/cli.bundle.mjs vaults
node dist/cli/cli.bundle.mjs vaults --chain eip155:1
node dist/cli/cli.bundle.mjs vaults --sort apy --order desc --page 1 --page-size 10
node dist/cli/cli.bundle.mjs vaults --assets 0x8d0d...,0x55d3... --curators "Lista DAO,Pangolins"

4) `markets`

Purpose: Discover markets from SDK list API.

Common options:

--page, --page-size, --sort, --order
--zone, --keyword
--loans <a,b>, --collaterals <a,b>

Examples:

bash
node dist/cli/cli.bundle.mjs markets
node dist/cli/cli.bundle.mjs markets --chain eip155:56 --sort liquidity --order desc --page-size 20
node dist/cli/cli.bundle.mjs markets --loans USD1,USDT --collaterals USD1,BTCB

Notes:

Command filters out SmartLending (zone=3) and fixed-term (termType=1) markets in output.
Always include this user-facing note when presenting market list:
Smart Lending and fixed-term market operations are currently not supported in this skill. For full functionality, please use the Lista website.

5) `holdings`

Purpose: Query positions by wallet address.

Options:

--address <0x...> (optional if context already has userAddress)
--scope <all|vault|market|selected>

Examples:

bash
# all positions (vault + market)
node dist/cli/cli.bundle.mjs holdings --address 0xYOUR_ADDRESS

# only vault positions
node dist/cli/cli.bundle.mjs holdings --address 0xYOUR_ADDRESS --scope vault

# only market positions
node dist/cli/cli.bundle.mjs holdings --address 0xYOUR_ADDRESS --scope market

# only currently selected position
node dist/cli/cli.bundle.mjs holdings --scope selected

Market position fields include:

User-facing fields:
isSmartLending (zone === 3)
isFixedTerm (termType === 1)
isActionable (!isSmartLending && !isFixedTerm)
Raw JSON technical fields (debug/integration only):
zone, termType

6) `select`

Purpose: Persist active target in context for follow-up operations.

Modes:

select vault: --vault
select market: --market
read context: --show
clear context: --clear

Examples:

bash
# select vault
node dist/cli/cli.bundle.mjs select \
  --vault 0xfa27f172e0b6ebcef9c51abf817e2cb142fbe627 \
  --chain eip155:56 \
  --wallet-topic <topic> \
  --wallet-address 0xYOUR_ADDRESS

# select market
node dist/cli/cli.bundle.mjs select \
  --market 0xd384584abf6504425c9873f34a63372625d46cd1f2e79aeedc77475cacaca922 \
  --chain eip155:56 \
  --wallet-topic <topic> \
  --wallet-address 0xYOUR_ADDRESS

# show/clear
node dist/cli/cli.bundle.mjs select --show
node dist/cli/cli.bundle.mjs select --clear

Notes:

Market selection rejects SmartLending and fixed-term markets.
Context file: ~/.agent-wallet/lending-context.json

7) `deposit`

Purpose: Deposit vault asset.

Required:

--amount
plus either selected vault context or explicit --vault + wallet info

Example:

bash
# using selected vault
node dist/cli/cli.bundle.mjs deposit --amount 1

# explicit target
node dist/cli/cli.bundle.mjs deposit \
  --vault 0xVAULT \
  --amount 1 \
  --chain eip155:56 \
  --wallet-topic <topic> \
  --wallet-address 0xYOUR_ADDRESS

8) `withdraw`

Purpose: Withdraw vault asset.

Required:

one of --amount or --withdraw-all

Examples:

bash
node dist/cli/cli.bundle.mjs withdraw --amount 0.5
node dist/cli/cli.bundle.mjs withdraw --withdraw-all

9) `supply`

Purpose: Supply market collateral.

Required:

--amount
plus selected market context or explicit market/wallet parameters

Example:

bash
node dist/cli/cli.bundle.mjs supply --amount 2

10) `borrow`

Purpose: Borrow loan token, or simulate borrow capacity.

Modes:

simulate only: --simulate
simulate with hypothetical supply: --simulate --simulate-supply <amt>
execute borrow: --amount <amt>

Examples:

bash
# check max borrowable
node dist/cli/cli.bundle.mjs borrow --simulate

# check max after hypothetical supply
node dist/cli/cli.bundle.mjs borrow --simulate --simulate-supply 2

# execute borrow
node dist/cli/cli.bundle.mjs borrow --amount 0.01

11) `repay`

Purpose: Repay market debt, or simulate repay impact.

Modes:

simulate repay amount: --simulate --amount <amt>
simulate repay-all: --simulate --repay-all
execute repay: --amount <amt> or --repay-all

Examples:

bash
# simulate partial repay impact
node dist/cli/cli.bundle.mjs repay --simulate --amount 0.01

# simulate full repay impact
node dist/cli/cli.bundle.mjs repay --simulate --repay-all

# execute
node dist/cli/cli.bundle.mjs repay --amount 0.01
node dist/cli/cli.bundle.mjs repay --repay-all

12) `market-withdraw`

Purpose: Withdraw market collateral.

Required:

one of --amount or --withdraw-all

Examples:

bash
node dist/cli/cli.bundle.mjs market-withdraw --amount 0.5
node dist/cli/cli.bundle.mjs market-withdraw --withdraw-all

Transaction Behavior

Transactions are built by SDK and executed through lista-wallet-connect call command.
call performs simulation by default before requesting wallet signature.
If simulation fails, result is returned as error/reverted and no signing request is sent.
If user rejects in wallet, command returns status: "rejected", reason: "user_rejected".
For multi-step operations (approve + action), responses include completed step count and failed/pending step.

Typical Workflows

Vault flow

bash
# 1) discover
node dist/cli/cli.bundle.mjs vaults --chain eip155:56

# 2) select
node dist/cli/cli.bundle.mjs select --vault 0xVAULT --wallet-topic <topic> --wallet-address 0xADDR

# 3) operate
node dist/cli/cli.bundle.mjs deposit --amount 1
node dist/cli/cli.bundle.mjs withdraw --amount 0.5

Market flow

bash
# 1) discover
node dist/cli/cli.bundle.mjs markets --chain eip155:56

# 2) select
node dist/cli/cli.bundle.mjs select --market 0xMARKET --wallet-topic <topic> --wallet-address 0xADDR

# 3) operate
node dist/cli/cli.bundle.mjs supply --amount 2
node dist/cli/cli.bundle.mjs borrow --simulate
node dist/cli/cli.bundle.mjs borrow --amount 0.01
node dist/cli/cli.bundle.mjs repay --simulate --amount 0.01
node dist/cli/cli.bundle.mjs repay --amount 0.01
node dist/cli/cli.bundle.mjs market-withdraw --amount 1

Security Checklist

Confirm token symbol and amount with user before any state-changing action.
Confirm chain (eip155:56 vs eip155:1) before sending tx.
Prefer borrow --simulate before first borrow on a market.
Treat user_rejected as normal user decision, not protocol failure.

Illustrative Want to see what leoyeai/lista-lending would put in front of your model? Preview the rendered body and check its trigger phrases — no LLM call, no signup.

Latest benchmark · publisher runVerified

8/8/2026, 18:27 UTC · 25 cases

Measured on 9 comparable cases — below the 22 needed to publish a lift, so this result is shown here but not on the registry card or leaderboard.

Lift

+8%

36% vs 28%

Agent turns

1.0 vs 1.0

Tokens

+280%

5,358 vs 1,411.33

fail → pass · 3pass → pass · 6fail → fail · 15pass → fail · 1

Per-case results (25)

Case	Status	Duration (ms)			Turns			Tokens			Tool calls
Case	Status	Without	With	Δ	Without	With	Δ	Without	With	Δ	Without	With	Δ
case-01 Can you check all my active Lista lending holdings for wallet 0x71C7656EC7ab88b098defB751B7401B5f6d8976F on Binance Smart Chain? Please output a human-readable summary block followed by separate tables for my vault deposits and market borrow positions instead of printing raw CLI JSON.	fail→fail	7,770	9,460	+22%	1	1	0%	642	5,404	+742%	0	0	—
case-13 I want to check how much collateral capacity I would have if I added 5 tokens of supply before borrowing on my selected market.	fail→fail	8,072	8,974	+11%	1	1	0%	1,482	5,164	+248%	0	0	—
case-02 Please show me available Lista lending markets on BSC ranked by liquidity. Render the results in a formatted summary table listing collateral tokens, loan assets, borrow rates, and available liquidity without outputting raw internal payload objects.	fail→fail	8,176	7,006	-14%	1	1	0%	1,525	5,259	+245%	0	0	—
case-03 I'm considering borrowing USDT against my supplied collateral in my active Lista market context on chain eip155:56. First run a simulation to show my maximum borrow allowance and potential health factor impact, then guide me through borrowing 50 USDT with explicit user consent before submitting.	fail→fail	8,636	7,324	-15%	1	1	0%	1,486	5,286	+256%	0	0	—
case-04 I want to open a CDP position on Lista to mint lisUSD against my collateral. Can you execute the CDP deposit and mint transaction for me?	pass→pass	7,103	3,608	-49%	1	1	0%	1,195	5,327	+346%	0	0	—
case-05 I need a full daily digest and comprehensive risk report across all my Lista products, including collateral summaries and broad protocol analytics. Can you generate this report?	fail→pass	13,953	3,213	-77%	1	1	0%	2,212	5,365	+143%	0	0	—
case-06 My WalletConnect session dropped and I need to establish a new wallet pairing topic with my mobile wallet app before running commands. Can you run the pairing command?	pass→pass	4,304	3,876	-10%	1	1	0%	706	5,430	+669%	0	0	—
case-07 I want to set active context to market 0xd384584abf6504425c9873f34a63372625d46cd1f2e79aeedc77475cacaca922 which is a SmartLending zone 3 market on BSC. Go ahead and select it.	fail→fail	3,926	9,472	+141%	1	1	0%	703	5,423	+671%	0	0	—
case-08 I have a market position on Lista with an LLTV of 80% (0.80) and a current LTV of 70% (0.70). What is my calculated health score and health status category?	fail→pass	7,700	3,909	-49%	1	1	0%	1,449	5,486	+279%	0	0	—
case-09 Display my holdings on chain eip155:1 in a human-readable table format.	pass→fail	5,196	7,823	+51%	1	1	0%	700	5,307	+658%	0	0	—
case-10 I have two non-actionable market holdings: one has isSmartLending set to true, and the other has isFixedTerm set to true. How should their reasons be labeled in the user-facing unsupported markets table?	pass→pass	8,855	2,763	-69%	1	1	0%	1,565	5,321	+240%	0	0	—
case-11 I want to view available vaults sorted by APY in descending order, but I did not pass an explicit chain flag. Which default chain parameter should be queried?	fail→pass	3,823	2,525	-34%	1	1	0%	524	5,190	+890%	0	0	—
case-12 I want to deposit 100 USDT into my currently selected vault context on BSC. I already ran select earlier.	fail→fail	3,831	7,773	+103%	1	1	0%	610	5,191	+751%	0	0	—
case-14 I want to test the health impact of paying off all my outstanding debt in my active market context without executing the transaction yet.	fail→fail	13,050	6,556	-50%	1	1	0%	1,901	5,240	+176%	0	0	—
case-15 Find vaults on Ethereum (eip155:1) where assets are 0x8d0d1 and curators are Pangolins.	fail→fail	23,368	6,868	-71%	1	1	0%	4,144	5,153	+24%	0	0	—
case-16 Clear my custom RPC override for BSC (eip155:56) back to default settings.	fail→fail	5,344	6,771	+27%	1	1	0%	935	5,153	+451%	0	0	—
case-17 Set my active vault target to 0xfa27f172e0b6ebcef9c51abf817e2cb142fbe627 on BSC using my wallet address 0x1111111111111111111111111111111111111111 and wallet topic top_123.	fail→fail	6,891	11,018	+60%	1	1	0%	859	5,637	+556%	0	0	—
case-18 I want to withdraw all my supplied collateral from my selected market context on chain eip155:56.	fail→fail	9,719	7,900	-19%	1	1	0%	1,489	5,303	+256%	0	0	—
case-19 I was attempting a vault withdrawal, but the wallet returned status 'rejected' with reason 'user_rejected'. How should this response be handled?	pass→pass	10,176	6,024	-41%	1	1	0%	1,621	5,678	+250%	0	0	—
case-20 When running Lista lending CLI commands, where should machine-readable result payloads be routed versus errors?	pass→pass	10,354	1,664	-84%	1	1	0%	1,548	5,030	+225%	0	0	—
case-21 Show me what vault or market context is currently persisted in my agent wallet context file.	fail→fail	4,961	6,021	+21%	1	1	0%	806	5,106	+533%	0	0	—
case-22 Supply 10 collateral tokens into selected market context on chain eip155:56.	fail→fail	8,740	7,857	-10%	1	1	0%	1,522	5,272	+246%	0	0	—
case-23 When listing user holdings, should raw internal fields like zone and termType be displayed in standard user tables?	pass→pass	12,592	3,655	-71%	1	1	0%	1,882	5,395	+187%	0	0	—
case-24 Withdraw 2.5 tokens from my currently selected Lista vault context.	fail→fail	5,280	7,251	+37%	1	1	0%	671	5,272	+686%	0	0	—
case-25 I want to execute a repay of 15 USDT on my selected market after asking for my final consent.	fail→fail	3,915	6,072	+55%	1	1	0%	604	5,245	+768%	0	0	—

These cases were also screened by the skill’s intent / safety pass.See the Trust review ↗Reproduce this benchmark in Colab ↗

How this skill was measured

DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 25 cases were attempted, and 9 counted toward the lift figure. The other 16 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +8 percentage points is the difference between those two pass rates over the 9 comparable cases. 4 cases got worse with the skill loaded, and they are included in that figure.

Model measured: gemini-3.6-flash
Graded by: gemini-3.6-flash
Grading method: judged
Verification: verified
Pass rate without the skill: 28%
Pass rate with the skill: 36%
Cases attempted: 25
Cases counted toward lift: 9
Cases that regressed: 4
Skill version measured: v1
Run completed: 8/8/2026, 18:27 UTC

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-05

Prompt

I need a full daily digest and comprehensive risk report across all my Lista products, including collateral summaries and broad protocol analytics. Can you generate this report?

Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.

Show both answers

SkillScore blends 4 signals1 of 4 complete

BenchmarkPass rate WITH skill on a verified run36

Live evalNeeds ≥10 evaluated traces in 30 days

AI ratingNeeds ≥10 rated traces

AdoptionNeeds ≥10 serves across ≥2 orgs in 30 days

Efficiency · separate axis (at equal correctness)+0% turns+278% tokens

Shown for transparency; not credited unless a verified run shows no correctness regression.

Backed by 1 of 4 signals — each added signal makes the score more robust.

Verified on gemini-3.6-flash · re-tested 50 days ago

Other measured skills in the registry, with their headline benchmark lift.

Ready to supercharge your AI agents?

leoyeai/lista-lending

SkillSafety

Source

Lista Lending Skill (POC)

Agent Execution Policy

When to Use This Skill

Routing Priority (Required)

Scope

Temporary Limitations

Project Structure

Prerequisites

Setup (Agent-Run)

Output Contract

Agent Display Guidelines

Supported Chains

Command Index

Global/Shared Options

Command Details

1) version

2) config

3) vaults

4) markets

5) holdings

6) select

7) deposit

8) withdraw

9) supply

10) borrow

11) repay

12) market-withdraw

Transaction Behavior

Typical Workflows

Vault flow

Market flow

Security Checklist

How this skill was measured

Measured example — same prompt, with vs. without the skill · from the benchmark run · case case-05

Related skills

1) `version`

2) `config`

3) `vaults`

4) `markets`

5) `holdings`

6) `select`

7) `deposit`

8) `withdraw`

9) `supply`

10) `borrow`

11) `repay`

12) `market-withdraw`