okx · maixuancanh · May 16, 2026 · May 16, 2026
diff --git a/skills/safe-competition-trading-agent/.claude-plugin/plugin.json b/skills/safe-competition-trading-agent/.claude-plugin/plugin.json
@@ -0,0 +1,18 @@
+{
+  "name": "safe-competition-trading-agent",
+  "description": "Competition-aware Agentic Wallet trading skill with risk-gated execution",
+  "version": "1.0.0",
+  "author": {
+    "name": "Safe Competition Trading Agent Contributors"
+  },
+  "license": "MIT",
+  "keywords": [
+    "trading",
+    "competition",
+    "agentic-wallet",
+    "xlayer",
+    "solana",
+    "security",
+    "risk"
+  ]
+}
diff --git a/skills/safe-competition-trading-agent/.gitignore b/skills/safe-competition-trading-agent/.gitignore
@@ -0,0 +1,11 @@
+__pycache__/
+*.py[cod]
+.pytest_cache/
+.mypy_cache/
+.coverage
+dist/
+build/
+*.egg-info/
+.env
+.venv/
+venv/
diff --git a/skills/safe-competition-trading-agent/LICENSE b/skills/safe-competition-trading-agent/LICENSE
@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2026 Safe Competition Trading Agent Contributors
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.
diff --git a/skills/safe-competition-trading-agent/README.md b/skills/safe-competition-trading-agent/README.md
@@ -0,0 +1,94 @@
+# Safe Competition Trading Agent
+
+Competition-aware Agentic Wallet trading skill with risk-gated execution.
+
+This plugin is designed as the main strategy skill for OKX Agentic Wallet trading competitions. It uses onchainOS as the primary data and trading layer, then composes a safe workflow with `agent-workflow-composer` and blocks unsafe trades with `agent-risk-firewall`.
+
+## Why This Exists
+
+Agentic trading competitions need more than a simple swap command. A useful agent must know:
+
+- whether the competition is active;
+- whether the wallet has joined;
+- which chains count for the competition;
+- whether the selected pair is an eligible real-token trade;
+- whether quote, slippage, price impact, and transaction context are safe enough;
+- whether the user has explicitly confirmed execution.
+
+`safe-competition-trading-agent` turns those checks into an ordered strategy workflow.
+
+## What It Does
+
+| Area | Behavior |
+| --- | --- |
+| Competition preflight | Plans `onchainos competition list/detail/user-status` before trading. |
+| Candidate selection | Scores supplied onchainOS token candidates and prefers liquid, lower-risk real tokens. |
+| Quote and tx context | Requires onchainOS swap quote and unsigned transaction preparation before firewall. |
+| Workflow validation | Uses a deterministic plan that places composer and firewall before execution. |
+| Risk gate | Requires `agent-risk-firewall` verdict before execution. |
+| Execution safety | Defaults to dry-run; live execution requires confirmation and an allow/warn verdict. |
+
+## Commands
+
+```powershell
+safe-competition-trading-agent plan --input request.json --format json
+safe-competition-trading-agent dry-run --input request.json --format json
+safe-competition-trading-agent execute --input request.json --format json
+safe-competition-trading-agent validate --input plan.json --format json
+safe-competition-trading-agent template --name competition-safe-swap
+safe-competition-trading-agent self-test
+```
+
+## Example Request
+
+```json
+{
+  "intent": "Dry-run a safe competition trade with a 10 USD budget.",
+  "chain": "xlayer",
+  "goal": "safe-volume",
+  "budgetUsd": "10",
+  "tokenIn": {"symbol": "USDC"},
+  "executionMode": "dry-run"
+}
+```
+
+## Live Execution Boundary
+
+The CLI does not sign or broadcast transactions by itself. It returns a guarded execution command only when:
+
+- `executionMode` is `confirm-before-execute`;
+- `confirmed` is `true`;
+- risk verdict is `allow`, or `warn` with confirmation;
+- the workflow has passed all safety gates.
+
+The host agent still executes through onchainOS Agentic Wallet.
+
+## Testing
+
+From the repository root:
+
+```powershell
+$env:PYTHONDONTWRITEBYTECODE = "1"
+python -m pytest .\skills\safe-competition-trading-agent\tests -q -p no:cacheprovider
+& "$env:USERPROFILE\.local\bin\plugin-store.exe" lint .\skills\safe-competition-trading-agent
+```
+
+Expected results:
+
+```text
+tests passed
+Plugin 'safe-competition-trading-agent' passed all checks
+```
+
+## Safety Notes
+
+- Do not execute before competition detail and user-status are known.
+- Do not execute before quote and unsigned transaction context exist.
+- Do not execute before `agent-risk-firewall check`.
+- Do not execute on a `block` verdict.
+- Do not use `--force`.
+- Do not request, store, or export private keys, seed phrases, or mnemonics.
+
+## License
+
+MIT
diff --git a/skills/safe-competition-trading-agent/SKILL.md b/skills/safe-competition-trading-agent/SKILL.md
@@ -0,0 +1,239 @@
+---
+name: safe-competition-trading-agent
+description: "Competition-aware Agentic Wallet trading skill that uses onchainOS data, risk firewall checks, and explicit user confirmation before execution"
+version: "1.0.0"
+author: "Safe Competition Trading Agent Contributors"
+tags:
+  - trading
+  - competition
+  - agentic-wallet
+  - xlayer
+  - solana
+  - security
+  - risk
+---
+
+# Safe Competition Trading Agent
+
+## Overview
+
+Use this skill when the user wants an Agentic Wallet agent to plan or execute a safer competition trade. This is the main judged strategy skill: it uses onchainOS as the primary data source and trading tool, then uses `agent-workflow-composer` and `agent-risk-firewall` as safety layers.
+
+The skill is competition-aware. It checks whether the wallet is registered, whether the competition is active, which chains count, whether the candidate is a real token trade, and whether the quote/unsigned transaction passes a risk gate before the agent asks the user to confirm execution.
+
+Default mode is `dry-run`. Live execution is allowed only after:
+
+1. onchainOS competition detail and user-status are known.
+2. onchainOS quote and unsigned transaction context are prepared.
+3. `agent-risk-firewall check` returns `allow`, or returns `warn` and the user explicitly confirms.
+4. The user has explicitly requested execution.
+
+## Trigger Phrases
+
+Use this skill for requests like:
+
+- "trade for competition"
+- "safe competition trade"
+- "agentic trading competition"
+- "find a safe token for competition"
+- "trade 10 dollars for contest"
+- "optimize my competition rank"
+- "increase my competition volume"
+- "dry-run competition trade"
+- "run a safe competition swap"
+- "join competition and trade"
+- "check whether this competition trade counts"
+- "find an eligible token trade"
+- "prepare a guarded competition swap"
+- "dry-run my Agentic Trading contest trade"
+
+Do not use this skill for generic wallet balance, normal token portfolio checks, exporting wallets, or unrelated DeFi positions. Use the corresponding wallet, portfolio, DeFi, or DEX skills for those.
+
+## Required onchainOS Role
+
+This skill must use onchainOS as the primary source of truth:
+
+| Purpose | Required command family |
+|---|---|
+| Wallet session | `onchainos wallet status` |
+| Competition discovery and rules | `onchainos competition list`, `onchainos competition detail` |
+| Registration status | `onchainos competition user-status` |
+| Token discovery and metadata | `onchainos token ...` or the active onchainOS token skill |
+| Quote | `onchainos swap quote` |
+| Unsigned transaction context | onchainOS swap transaction preparation command; prepare only, do not sign or broadcast |
+| Live execution after gates | `onchainos swap execute` |
+
+Use `agent-workflow-composer` to validate workflow order and `agent-risk-firewall` to produce the final risk verdict.
+
+## Competition Invariants
+
+Apply these rules when building the plan:
+
+- Every active competition currently counts Solana plus the backend primary chain. If detail says X Layer, treat both X Layer and Solana as supported until backend exposes a full multi-chain field.
+- Trades on either supported chain count toward the same standing.
+- Keep internal ids such as `activityId`, `chainIndex`, and `accountId` in tool context only. Do not render them in user-visible messages.
+- Identify competitions to the user by name, not by numeric id.
+- `myRankInfo.userTotal = 0` can mean the user has not hit the threshold or the backend has not updated yet; it does not mean the chain is unsupported.
+
+## CLI Commands
+
+```bash
+safe-competition-trading-agent plan --input request.json --format json
+safe-competition-trading-agent dry-run --input request.json --format json
+safe-competition-trading-agent execute --input request.json --format json
+safe-competition-trading-agent validate --input plan.json --format json
+safe-competition-trading-agent template --name competition-safe-swap
+safe-competition-trading-agent self-test
+```
+
+The CLI never signs or broadcasts. It creates plans, dry-run decisions, guarded execution commands, and local validation reports.
+
+## Input Contract
+
+```json
+{
+  "intent": "Dry-run a safe competition trade with a 10 USD budget.",
+  "chain": "xlayer",
+  "goal": "safe-volume",
+  "budgetUsd": "10",
+  "tokenIn": {"symbol": "USDC"},
+  "executionMode": "dry-run",
+  "confirmed": false,
+  "competition": {
+    "activityName": "Selected Agentic Trading competition",
+    "active": true,
+    "joined": true,
+    "chainName": "X Layer",
+    "supportedChains": ["xlayer", "solana"],
+    "eligibleTokenTradeRequired": true
+  },
+  "candidates": [
+    {
+      "symbol": "MEME",
+      "address": "0x...",
+      "liquidityUsd": "150000",
+      "volume24hUsd": "75000",
+      "riskLevel": "LOW"
+    }
+  ],
+  "quote": {
+    "expectedOut": "12345",
+    "slippagePct": 1,
+    "priceImpactPct": 1
+  },
+  "riskVerdict": {
+    "verdict": "allow",
+    "reasons": []
+  }
+}
+```
+
+Supported values:
+
+- `chain`: `xlayer`, `solana`
+- `goal`: `safe-volume`, `rank-optimizer`, `eligible-token`, `custom`
+- `executionMode`: `dry-run`, `confirm-before-execute`
+
+## Execution Flow
+
+Follow this exact order:
+
+1. Parse user intent: chain, budget, goal, dry-run vs execution.
+2. Run `onchainos wallet status`.
+3. Run `onchainos competition list --status 0`.
+4. Select the competition by user intent or ask the user to choose by name.
+5. Run `onchainos competition detail --activity-id <activityId>`.
+6. Run `onchainos competition user-status --activity-id <activityId> --evm-wallet <evmWallet> --sol-wallet <solWallet>`.
+7. If the user has not joined, warn that the trade may not count. Ask whether to join before trading.
+8. Query onchainOS token data for candidate tokens on the selected chain.
+9. Exclude stable/native-only candidates for competition mode unless the user explicitly asks for a non-eligible dry-run.
+10. Run `onchainos swap quote`.
+11. Prepare unsigned swap transaction context through onchainOS only after the user has approved the dry-run preparation step. This preparation step is not live execution and must not sign or broadcast.
+12. Run `agent-workflow-composer plan` with `workflowType: "competition-trade"`.
+13. Run `agent-risk-firewall check` with `policyProfile: "competition"`.
+14. Apply the verdict:
+    - `block`: stop. Do not ask for signature.
+    - `warn`: show reasons and ask for explicit confirmation.
+    - `allow`: continue only if the user already requested execution.
+15. If execution is confirmed, run `onchainos swap execute`.
+16. Show tx result, risk decision id, and next step.
+
+## Output Format
+
+For user-facing output, use this structure:
+
+```md
+## Competition Trade Plan
+
+Competition: <name only, no internal id>
+Chain: <X Layer or Solana>
+Mode: <dry-run or confirm-before-execute>
+Budget: <amount and token>
+
+## Candidate
+
+Token: <symbol>
+Reason:
+- <selection reason>
+
+## Quote
+
+Route: <route>
+Estimated out: <amount>
+Slippage: <percent>
+Price impact: <percent>
+
+## Risk Verdict
+
+Verdict: <allow | warn | block>
+Risk score: <0-100>
+Reasons:
+- <reason code and short message>
+
+## Next Step
+
+<continue, ask confirmation, join competition first, or cancel>
+```
+
+Do not show `activityId`, `accountId`, private keys, seed phrases, mnemonics, raw secrets, or internal-only identifiers.
+
+## Safety Rules
+
+- Default to `dry-run`.
+- Never execute when `agent-risk-firewall` returns `block`.
+- Never execute on `warn` without explicit user confirmation.
+- Never execute if `competition_user_status` is missing.
+- Never execute if quote or unsigned transaction context is missing.
+- Never use `--force`.
+- Never export or request private keys, seed phrases, or mnemonics.
+- Never claim that a trade will definitely win a competition.
+
+## Error Handling
+
+| Situation | Response |
+|---|---|
+| Wallet not logged in | Ask user to run or approve wallet login before trading. |
+| No active competition | Stop and show that no active competition was found. |
+| User not joined | Warn and ask whether to join before trading. |
+| No token candidate | Query onchainOS token/signal data; if still none, stop. |
+| Quote unavailable | Retry once; if still unavailable, stop. |
+| Firewall unavailable | Treat as `warn` or stop in strict review contexts; do not execute blindly. |
+| Firewall block | Cancel trade. |
+
+## Example Agent Behavior
+
+User: "Trade 10 dollars for a safe competition token on X Layer."
+
+Agent should:
+
+1. Explain it will dry-run first.
+2. Fetch wallet and competition context through onchainOS.
+3. Select a real token candidate from onchainOS data.
+4. Quote and prepare unsigned tx context through onchainOS.
+5. Run workflow composer and risk firewall.
+6. Show the structured output.
+7. Ask for explicit confirmation only if the trade is not blocked.
+
+## Disclaimer
+
+This skill is a trading guardrail and strategy workflow, not a guarantee of profit or safety. On-chain data, competition rules, rankings, quotes, and simulations can change or be stale. Trading can cause loss of funds.