version: "1.0.1" name: okx-growth-competition description: "Agentic Wallet exclusive trading competitions. Full lifecycle: discover → view rules → join → trade → check rank → claim reward. Triggers: 'list trading competitions', 'show available competitions', 'view competition details', 'show competition rules', 'show prize pool', 'register for competition', 'join trading contest', 'check my competition status', 'view leaderboard', 'check my ranking', 'claim competition reward', 'did I win', 'winners list', 'show registered wallet', 'export wallet'. Do NOT use for: general DEX swaps (use okx-dex-swap); portfolio / PnL queries outside a competition (use okx-wallet-portfolio or okx-dex-market); wallet login or tx history (use okx-agentic-wallet); any non-competition trading activity. Do NOT use when the user says only a single word like 'competition', 'rank', or 'claim' without naming a specific competition." license: MIT metadata: author: okx version: "1.2.0" homepage: "https://web3.okx.com"

OKX Growth Competition — Trading Competition

Agentic Wallet exclusive trading competitions. Full lifecycle: list → detail → join → trade → rank → claim.

CLI reference: references/cli-reference.md

Facts about every Agentic Wallet competition (always true unless backend confirms otherwise)

<MUST> Treat the following as factual ground truth when the user asks about how a competition works. Every activity exposes two chain-related fields, and both contribute to the trading set:

• chainId — single id. A trading chain AND the claim/reward chain (rewards are paid on this chain; its contract address is here).
• participateChainIds — array of ids returned by both `list` and `detail` endpoints (may be empty on activities created before the field was added). Additional trading chains (no claim role).

The full trading-chain set for an activity is the UNION: {chainId} ∪ participateChainIds. Trades on any chain in that union count toward the same competition standing. The claim chain is chainId only. </MUST>

1. The trading-chain set is `{chainId} ∪ participateChainIds` (dedup). Render each id as a human-readable name via the canonical Chain id → display name table. Currently supported competition chains: 1 → Ethereum, 196 → X Layer, 501 → Solana.
2. Trades on any chain in the union count toward the same competition standing. Trades on chains NOT in the union do not count. Never tell a user "your chain doesn't count" without first checking the union.
3. myRankInfo.userTotal = 0 means the user has not yet hit the qualifying threshold or the backend metric pipeline has not picked up their trades yet — it does NOT mean the user's chain is unsupported.
4. competition_rank takes a single optional wallet. Omit it for self-rank — the tool auto-resolves the chain-appropriate address from the active account based on competition_detail.chainId (the primary chain used for reward claim and rank indexing). Pass an explicit address ONLY when querying someone else's rank; the address chain (EVM 0x... else Solana) must match the activity's primary chain or the tool rejects the call (no silent wrong-chain queries). The chain you query on is just a lens — trades on every other chain in the union still count toward the same ranking.
5. Claim path uses only `chainId` (that's where the reward contract / activity address lives). Everything else — trading eligibility, "which chains count", chain-list display in templates — uses the union of both fields.

When the user asks "Does my chain count for this competition?" or "Which chains can I trade on?", answer from the union of chainId and participateChainIds. When rendering the "Chain" column / line in a template, render that same union (deduplicated) — see the {supportedChains} computation rule in each Step's field-mapping section.

Identity resolution invariant (deterministic — always answerable)

<MUST> The query identity for competition_rank and competition_user_status is mutually exclusive: the backend accepts EITHER accountId (self-query, multi-chain by design) OR walletAddress (cross-user query on a specific chain) — never both, never neither. When the user asks "which identity did you use?", the answer is deterministic from the call shape — never reply with "both", "unknown", or "I'm not sure". </MUST>

Algorithm (always runs internally; AI cannot bypass):

How to answer "which identity did you use?":

Forbidden answers (paraphrase patterns observed and are wrong):

• ❌ "The tool sends both accountId and walletAddress" — Wrong. Exactly one, never both.
• ❌ "The tool picks EVM or SOL based on chain" — Wrong (post-refactor). That was the old wallet-only model. Self-queries now send accountId, no chain pick.
• ❌ "I'm not sure which one the tool ended up sending" — Wrong. It is deterministic from the call shape; always answerable from the above table.

For multi-activity queries (e.g. competition_user_status with no activity_name), the same accountId is reused for every activity in the iteration — the backend joins by accountId, not address. The answer is uniformly "I used your accountId for all activities".

⚠️ Mandatory reading order

<MUST> Before producing ANY user-facing message about a competition (list / detail / join / claim / rank / status / wallet-export-guard), you MUST first locate the matching `Step N` section below and follow its fixed template structure. Do NOT improvise the format. Do NOT shorten the templates. Do NOT drop sections or merge them.

The template structure is fixed; the language follows the user — see the ## Output Language rule above. When the user writes Chinese, translate the template strings to natural Chinese. When the user writes English, use English as written. Placeholders (including chain display names from {supportedChains}) stay as-is.

Quick router (user intent → template section):

• "list competitions / show available competitions" → Step 1 (table, optionally split by Active / Ended)
• "show details / show rules / show prize pool" → Step 2 (Basic info block + 4 reward sections, with {supportedChains} chain line and required participation / Skill copy)
• "register / join" → Step 3 (registration success fixed template + disclaimer)
• "trade for me" → Step 4 (delegate to okx-dex-swap)
• "leaderboard / ranking" → Step 5
• "claim reward" → Step 6 (use competition_claim MCP, atomic)
• "show registered wallet" → Additional Flows / Query Registered Wallet
• "export wallet" → Additional Flows / Wallet Export Guard

If the user's intent does not clearly map to one of the above, ask which they meant before responding — do not invent a freeform format. </MUST>

Pre-flight

Read ../okx-agentic-wallet/_shared/preflight.md. If missing, read _shared/preflight.md.

Command Index

--status (request filter): 0=active, 1=ended, 2=all activityStatus (response field): `3`=active, `4`=ended — these are DIFFERENT values from the request filter sort-type: dynamic — read from competition_detail → tabConfigs[].rankFieldConfig[].sortValueMap.descend. Currently observed values: 1=PnL% (realized ROI), 7=PnL (realized profit). Future activities may add more — always trust tabConfigs over hardcoding.

Output Rules

Internal-only IDs vs user-facing display. Internal numeric IDs (activityId, chainIndex, accountId) are returned in tool responses on purpose — they are needed to chain calls between tools (e.g. after competition_join, you may need to call competition_detail with the activity id to fill the success template). Keep them in the data layer; never render them in user-visible messages.

<NEVER> Never include any internal id in a message produced for the user — under ANY circumstance, in ANY format. Identify activities to the user EXCLUSIVELY by activityName (or shortName if name is unavailable). </NEVER>

Forbidden user-visible patterns (do NOT produce output like this):

• ❌ Agentic Trading Contest (#107)
• ❌ #106 (agenticwallettest1)
• ❌ A column titled ID, Activity ID, #, or any equivalent in the user's language
• ❌ A row labeled Activity ID, ID, #, or any equivalent in the user's language (e.g. a 2-column key/value table where a row maps the id label to the numeric value) — this is the same violation as a column, just rotated
• ❌ Any reference like competition 107, id 107, the activity id is 107, or the same wording in another language
• ❌ Numeric id rendered anywhere a user can see it, regardless of label, shape, or language

Correct user-visible pattern:

• ✅ Agentic Trading Contest
• ✅ When disambiguating two activities with the same name, append chainName (e.g. Agentic Trading Contest (Solana)), never the ID.

Behind the scenes (allowed and expected):

• ✅ Reading activityId from a competition_user_status / competition_join response and passing it to competition_detail to fetch the data needed by a fixed template.
• ✅ Any tool-to-tool chaining via numeric ids — as long as the final user-facing message omits them.

When the user asks to act on a specific activity (e.g. "claim Agentic Trading Contest"), the MCP tools competition_claim / competition_join accept activity_name and resolve the id server-side, so you can also use names directly without doing your own lookup.

Time Formatting (MANDATORY)

<MUST> All timestamps from competition APIs MUST be rendered using these exact rules. Do NOT free-style time conversions. </MUST>

Preferred — use backend-formatted strings when available:

For the competition_detail response, the backend now returns pre-formatted UTC+8 strings — use them verbatim, do NOT recompute from the raw epoch:

When rendering, take the string as-is and append (UTC+8) for the timezone marker (e.g. 2026-05-07 18:00:00 (UTC+8)). This is the only correct path for detail view times — the backend has already done the math; computing from startTime / endTime epoch invites off-by-an-hour errors and AI hallucination.

Raw fields — only use when no `*Formatted` counterpart is present:

If you see a 13-digit value where a 10-digit is documented (or vice versa), do NOT silently coerce — flag it as a backend anomaly.

Display format — exact, no improvisation:

• Detail view (with *Formatted available): YYYY-MM-DD HH:mm:ss (UTC+8) — example: 2026-05-07 18:00:00 (UTC+8) (use the formatted string + suffix)
• List view compact range: YYYY-MM-DD ~ YYYY-MM-DD (UTC+8 day boundary) — example: 2026-05-07 ~ 2026-05-21 (compute from raw startTime / endTime, take the date portion only)
• joinTime / claimTime in user-facing context: YYYY-MM-DD HH:mm (UTC+8) (compute from raw epoch)
• rankUpdateTime (last refresh marker): YYYY-MM-DD HH:mm:ss (UTC+8) (compute from raw ms epoch)

Timezone rule: ALL competition times displayed to the user are in UTC+8 (China Standard Time). The competition product is operated in UTC+8; never display raw UTC, never use the user's local timezone.

When you DO need to convert from raw epoch (no formatted field available):

1. Identify the field's documented unit (seconds or milliseconds — see table above).
2. Convert to a Date object using the correct unit.
3. Format as a UTC+8 wall-clock string per the display format above.
4. Always append the (UTC+8) suffix.

<NEVER>

• ❌ Do NOT recompute time from startTime / endTime epoch when startTimeFormatted / endTimeFormatted is present in the response — use the backend-formatted string directly.
• ❌ Do NOT shell out to date -r <epoch> (or any equivalent) when a *Formatted field is available — it adds tool-call noise for no benefit and risks platform-specific conversion bugs.
• ❌ Do NOT mentally compute the date from a raw timestamp using your training-data sense of "current date" — always either use the backend *Formatted field, or do the explicit numeric conversion.
• ❌ Do NOT mix seconds and milliseconds — a 13-digit value treated as seconds lands in year ~58000; a 10-digit value treated as ms lands in 1970.
• ❌ Do NOT drop the (UTC+8) suffix on date-time strings.
• ❌ Do NOT use the user's local timezone — even if the user is overseas, competition times are operated in UTC+8.

</NEVER>

Output Language

<MUST> Render every fixed template in the user's conversation language. The template structure (sections, ordering, numbered items, table column count, placeholder positions, the {supportedChains} placeholder, and the [Disclaimer: ...] block) is fixed and must NOT change. Only the natural-language text inside is translated to the user's language naturally.

Placeholders are never translated. {supportedChains}, {chainName}, {rewardUnit}, {txHash}, {accountName}, etc. are filled with API values verbatim — do not localize them. Chain display names (e.g. Solana, X Layer, Base) come from the canonical id → name mapping and stay as-is in every language. </MUST>

Execution Flow

Step 1 — Discover Competitions

Choosing the status filter

Decide which status to pass based on the user's intent:

When in doubt, prefer status=2 so the user can see the full picture and pick.

<MUST> Display the result as markdown tables — one row per competition. Do not use a numbered prose list, do not collapse fields into a single sentence.

When the result contains BOTH active (activityStatus=3) and ended (activityStatus=4) entries, split into two separate tables under bold subheadings (`Active` / `Ended`, translated to the user's language), in that order. When only one status is present, render a single table without a subheading. </MUST>

Fixed table template (English canonical; translate cells when user is non-English)

**Active**
 
| Name | Chain | Time | Total Prize Pool | Details |
|------|-------|------|------------------|---------|
| {name} | {supportedChains} | {startTime} ~ {endTime} | {rewards} | [View](https://web3.okx.com/boost/trading-competition/{shortName}) |
| ... | ... | ... | ... | ... |
 
**Ended**
 
| Name | Chain | Time | Total Prize Pool | Details |
|------|-------|------|------------------|---------|
| {name} | {supportedChains} | {startTime} ~ {endTime} | {rewards} | [View](https://web3.okx.com/boost/trading-competition/{shortName}) |
| ... | ... | ... | ... | ... |

For non-English users, translate the column headers, section headers, and link text naturally. The structure (column count, ordering, {supportedChains} placeholder) does not change.

Field-mapping rules

• Group rows by availableCompetitions[].status: 3 → Active table, 4 → Ended table.
• Name column ← name
• Chain column ← {supportedChains}, computed as the union of `participateChainIds` and `chainId`:

1. Start with the ids in participateChainIds (in backend-returned order).
2. If chainId is not already in that list, append it at the end.
3. Map each id to its display name. Currently supported competition chains: 1 → Ethereum, 196 → X Layer, 501 → Solana.
4. Join with , .

• Examples:
• chainId=196, participateChainIds=[501] → Solana, X Layer
• chainId=196, participateChainIds=[196, 501] → X Layer, Solana (chainId already in list — no duplicate)
• chainId=501, participateChainIds=[501] → Solana
• chainId=196, participateChainIds empty/missing (legacy activity created before the field was added) → X Layer (chainId only)
• Rationale: Both chainId and participateChainIds are trading chains — trades on any of them count. chainId additionally is the claim chain. The display union exposes the user to the full set so they can pick where to trade.
• Time column ← startTime ~ endTime formatted per Time Formatting rules above. List-table compact form: YYYY-MM-DD ~ YYYY-MM-DD in UTC+8 (e.g. 2026-05-07 ~ 2026-05-21). Do NOT include time-of-day in the compact list to keep the column narrow — full time-of-day is shown in Step 2 detail view only.
• Total Prize Pool column ← rewards field (already a formatted string like 50,000 USDC)
• Details column ← https://web3.okx.com/boost/trading-competition/<shortName> as a markdown link

After the table(s), ask the user (in their language):

• If only Active has entries: Which competition would you like to view in detail, or would you like to register directly?
• If only Ended has entries: Would you like to check your ranking or claim status for any of these?
• If both: combine — Which active competition would you like to register or view, or which ended competition would you like to check your ranking / claim?

Empty-result handling (English canonical; translate to user's language)

• All filters returned 0 entries → No trading competitions available right now.
• status=0 filter returned 0 entries → No active trading competitions at the moment.
• status=1 filter returned 0 entries → No ended trading competitions yet.

CLI equivalent

onchainos competition list --status 2   # all
onchainos competition list --status 0   # active only
onchainos competition list --status 1   # ended only

Step 2 — View Details (if requested)

onchainos competition detail --activity-id <id>

<MUST> Display competition / reward info using the fixed English template below. The structure (sections, ordering, numbered list, placeholder positions, the {supportedChains} placeholder on the chain line) is fixed. Copy the template character-for-character; only fill in placeholders. Do not paraphrase, abbreviate, or substitute synonyms.

When the user's language is not English, translate the natural-language strings to the user's language while preserving the structure, the placeholders, and every required content invariant listed below. Do not reorder, omit, or merge sections. </MUST>

Fixed display template

Basic Information
Supported chains: {supportedChains}
Duration: {startTime} ~ {endTime}
Total Prize Pool: {totalPrizePool}
 
Prize Categories:
Realized PNL% Prize Pool ({roiPoolAmount})
Ranked from highest to lowest by realized PNL%.
{roiRankTable}
 
Realized PnL Prize Pool ({pnlPoolAmount})
Ranked from highest to lowest by realized PNL amount.
{pnlRankTable}
 
Participation Prize ({participationPoolAmount})
Registered users who accumulate $100 or more in total trading volume via Agentic Wallet and maintain a total wallet balance of $100 or above throughout the competition period, will share the {participationPoolAmount} participation prize pool equally. Random asset snapshots will be taken during the competition period to verify eligibility.
 
Skill Quality Prize ({skillPoolAmount})
The Skill Quality Prize is an independently judged award. During the competition period, participants may submit their Agent Skills through the event landing page. Eligible submissions include, but are not limited to, on-chain autonomous yield strategies, trading analysis, and trading signal monitoring. All submitted Agent Skills will be evaluated through a dual-review process combining AI pre-screening and manual judging. The top {skillTopN} Skill creators by score will each receive a reward of {skillPerCreatorReward}.

Field-mapping rules

• Chain line ← {supportedChains}, computed as the union of `data.participateChainIds` and `data.chainId`:

1. Start with data.participateChainIds (in backend-returned order).
2. Append data.chainId at the end if not already present.
3. Map each id to its display name. Currently supported competition chains: 1 → Ethereum, 196 → X Layer, 501 → Solana.
4. Join with , .

• Examples (using real backend shapes):
• chainId=196, participateChainIds=[501] → Solana, X Layer
• chainId=196, participateChainIds=[196, 501] → X Layer, Solana
• chainId=501, participateChainIds=[501] → Solana
• chainId=196, participateChainIds empty/missing (legacy activity) → X Layer (chainId only)
• Both fields are trading chains (trades on any of them count toward the competition standing); chainId additionally hosts the reward contract / claim path. Display the union so the user sees the full trading set.
• {startTime} / {endTime} ← read startTimeFormatted / endTimeFormatted directly from competition_detail.data and append (UTC+8). Final form: YYYY-MM-DD HH:mm:ss (UTC+8) (e.g. 2026-05-07 18:00:00 (UTC+8)). Do NOT compute from raw startTime / endTime epoch — the backend has already done the math.
• {totalPrizePool} ← sum of all prizePoolDistribution[].totalReward plus rewardUnit (e.g. 50,000 USDC).
• {roiPoolAmount} ← totalReward of the realized-ROI tab.
• {pnlPoolAmount} ← totalReward of the realized-PnL tab.
• {participationPoolAmount} ← totalReward of the participation prize tab.
• {skillPoolAmount} ← totalReward of the Skill quality prize tab.
• {skillTopN} ← upper bound of the Skill tab's rules[].interval (e.g. "1-10" → 10).
• {skillPerCreatorReward} ← that rule entry's reward + rewardUnit (e.g. 500 USDC).
• {roiRankTable} / {pnlRankTable} ← markdown table built from the corresponding tab's rules[]. Format (English canonical; localize headers to user's language):

```

```

Interval / reward formatting per row:

• Single rank (interval = "1") → Rank cell Rank 1, Reward cell <reward> <rewardUnit> (no each prefix)
• Range (interval = "2-6") → Rank cell Ranks 2-6, Reward cell <reward> <rewardUnit> each
• Always end with a totals row whose Reward cell is the tab's totalReward + rewardUnit.

If any of the four pools is absent for a particular activity, omit just that section (keep the others as-is).

Required content invariants (per section)

Section 1 — Realized PNL% Prize Pool

• Title MUST be exactly Realized PNL% Prize Pool (or its faithful translation in the user's language). Do NOT substitute with PnL% Ranking Award / ROI Ranking Award / Realized ROI Pool.
• Description MUST mention: ranking by realized PNL%, highest to lowest.
• Rank table MUST have headers Rank / Reward and end with a Total row.

Section 2 — Realized PnL Prize Pool

• Title MUST be exactly Realized PnL Prize Pool. Do NOT substitute with PnL Ranking Award / Realized PnL Pool.
• Description MUST mention: ranking by realized PNL amount, highest to lowest.
• Rank table MUST follow the same format as Section 1.

Section 3 — Participation Prize (PRODUCT-MANDATED COPY)

• Title MUST be exactly Participation Prize.
• The description body MUST include all of these specific terms:
• Agentic Wallet
• accumulate $100 or more in total trading volume
• maintain a total wallet balance of $100 or above throughout the competition period
• share the participation prize pool equally
• random asset snapshots to verify eligibility

Section 4 — Skill Quality Prize (PRODUCT-MANDATED COPY)

• Title MUST be exactly Skill Quality Prize.
• The description body MUST include all of these specific terms:
• independently judged award
• submission of Agent Skills through the event landing page
• examples of eligible submissions (on-chain autonomous yield strategies, trading analysis, trading signal monitoring)
• dual-review process combining AI pre-screening and manual judging
• top {skillTopN} Skill creators ... each receive a reward of {skillPerCreatorReward}

<NEVER>

• ❌ Do NOT invent or omit chains on the chain line — {supportedChains} must reflect the union of `participateChainIds` and `chainId` (dedup, participateChainIds order first, then chainId if missing). Never drop chainId because participateChainIds is present; never drop participateChainIds because chainId exists.
• ❌ Do NOT reorder or merge the four reward sections — they must appear in the order 1 → 2 → 3 → 4.
• ❌ Do NOT add ID columns or expose any internal numeric id (activityId, etc.) anywhere in the output.
• ❌ Do NOT paraphrase, abbreviate, or substitute synonyms in Sections 3 and 4. These are product-mandated copy.
• ❌ Do NOT invent rank-distribution rules from the pool amount. The actual rules come from prizePoolDistribution[].rules[] — read them; do not divide.
• ❌ Do NOT use bullet markers (-) inside the four numbered sections — the structure is 1. Title (amount)\n description text then the rank table; not a bullet list.

</NEVER>

After printing the template, ask: Would you like me to register you for this competition?

Step 3 — Join (requires wallet login)

MCP: call competition_join with activity_name and chain_index only — evm_wallet and sol_wallet are auto-resolved from the active account.

CLI: pass addresses explicitly:

onchainos competition join --activity-id <id> --evm-wallet <evm_addr> --sol-wallet <sol_addr> --chain-index <chain_id>

Get chainIndex from competition_detail → chainIndex field.

If the user is not logged in, the tool returns not logged in — please run: onchainos wallet login. Tell the user verbatim:

Please run onchainos wallet login <your_email> in your terminal to log in (it cannot be done from inside this conversation), then ask me to register again.

Required pre-flight: distinguish duplicate-registration scenarios

<MUST> Before calling `competition_join`, you MUST first call `competition_user_status` for the activity to read the current account's `joinStatus`. This separates the two duplicate-registration cases that have different user-facing messages. </MUST>

Scenario A — current wallet already registered

Template:

Your current wallet account [accountName] is already registered for [activityName]. No need to register again. Would you like me to walk you through the rules in detail, or start trading directly?

Field-mapping:

• [accountName] ← accountName of the currently selected account (read from wallet_store / wallet status, e.g. Account 1)
• [activityName] ← activityName from the prior competition_user_status / competition_list response

Scenario B — same login, different account already registered

Triggered when competition_join returns code=11016 Participation limit reached.

Template:

Registration failed. Your wallet account [registeredAccountName] is already registered. You cannot register again. Please switch to your registered account to trade.

Field-mapping:

• [registeredAccountName] ← name of the OTHER account in the same login that holds the registration. To find it, iterate every account from wallet_store other than the current one and call competition_user_status for the activity, picking the one whose joinStatus=1. If no account is found (rare race), fall back to a generic phrase like another of your wallet accounts is already registered and ask the user to check onchainos wallet status themselves.

Successful registration

<MUST> On every successful `competition_join` call (the tool returns `joined: true`), output the fixed template below. Structure (the lead phrase + the supported-chains sentence + the closing question + the bracketed disclaimer on its own line) is fixed. {supportedChains} is the union of participateChainIds and chainId (see Field-mapping rules below); {totalPrizePool} is filled from competition_detail (call it before formatting if you don't have it cached). Translate the natural-language strings to the user's language while preserving structure and placeholders. </MUST>

Template:

Registered successfully! This competition runs on {supportedChains}, with a total prize pool of {totalPrizePool}. The trading contest ranks players by both PnL% and realized PnL, with additional Participation and Skill Quality Prizes. Would you like me to walk you through the detailed rules, or help you initiate a trade on {supportedChains}?
 
[Disclaimer: Digital asset trading involves risk. Prices can be highly volatile. Please understand the risks fully and do your own research before trading.]

Field-mapping rules

• {supportedChains} ← computed as the union of `data.participateChainIds` and `data.chainId` from competition_detail. Take participateChainIds in backend order, append chainId at the end if not already in the list, map each id to its display name (currently supported competition chains: 1 → Ethereum, 196 → X Layer, 501 → Solana), join with , . Examples: chainId=196+participateChainIds=[501] → Solana, X Layer; chainId=501+participateChainIds=[501] → Solana. The lead sentence and the closing question both use the same string; do not paraphrase by listing chains separately.
• {totalPrizePool} ← total reward pool (sum of all prizePoolDistribution[].totalReward + rewardUnit, e.g. 500 DJT).

<NEVER>

• ❌ Do NOT invent or omit chains on the chain line — {supportedChains} must reflect the union of `participateChainIds` and `chainId` (dedup, participateChainIds order first, then chainId if missing). Never drop chainId because participateChainIds is present; never drop participateChainIds because chainId exists.
• ❌ Do NOT drop or merge the four key phrases of the lead sentence: (1) which chains it runs on (from {supportedChains}), (2) the total prize pool, (3) the dual-axis PnL%/realized PnL ranking, (4) the existence of Participation and Skill Quality Prizes. These are required content; the wording can be localized but the four pieces must all appear.
• ❌ Do NOT drop the bracketed disclaimer line — it must appear on its own line at the end of the message, in the user's language.

</NEVER>

Other errors

On error containing `region` / `not available in your region`:

Registration failed: service is not available in your region. Please switch to a supported region and try again.

On any other error:

Operation failed. Please contact customer support.

Step 4 — Trade (delegate to okx-dex-swap)

When user asks to trade per competition rules:

Case A — User does NOT provide a CA (only token name/symbol):

1. Resolve the CA via the token_search MCP tool (CLI: onchainos token search).
2. Confirm with user before proceeding:

> Just to confirm, the CA for token "{tokenSymbol}" is "{contractAddress}". Is that correct?

1. Wait for user to confirm. Only proceed after explicit "yes".
2. Then follow Case B below.

Case B — User provides a CA directly:

1. Execute swap via the swap_swap MCP tool (CLI: onchainos swap swap); see the okx-dex-swap skill for parameters.
2. Report: "Done — your trade has been submitted." + tx hash.

Note: do NOT pre-empt the swap with an extra "token prices are volatile, do you accept the risk?" prompt. The user already requested the trade — additional friction is unwanted. Per-token risk metadata (e.g. honeypot / extreme volatility flags) belongs to okx-security and only fires when actually flagged.

Competition constraints per trade:

• Single-trade min $1 (orders below $1 are not counted)
• Token pairs must match competition rules from detail response

Step 5 — Check Status & Rank

Check participation status

onchainos competition user-status                       # all activities (uses accountId)
onchainos competition user-status --activity-id <id>    # single activity (uses accountId)

Display: join status, join time, reward status, reward amount.

• If rewardStatus=1 (won, not claimed): proactively ask "You have won a reward. Would you like me to claim it for you?"
• If rewardStatus=4 (pending draw): use the Pending-draw canonical template (English canonical below; translate to the user's language; substitute {activityName} from the activity's name / shortName field; do NOT paraphrase the 5-business-day window):

> "{activityName} has ended. The winners list is currently being finalized. The final reward list will be announced within 5 business days after the activity end — please return here to check your result and claim your reward then. Thank you for participating!"

• If rewardStatus=3 (expired): "Your reward has expired and can no longer be claimed."

Check leaderboard (full board)

<MUST> When the user says "view leaderboard" without specifying which one, you MUST:

1. Call competition_detail for the activity and enumerate tabConfigs[].rankFieldConfig[].sortValueMap.descend — this is the full set of leaderboards the activity exposes.
2. Call competition_rank ONCE PER sort_type (one HTTP call per leaderboard) so you have data for every leaderboard.
3. Render ALL of them in the response — one section per leaderboard. Do NOT silently default to a single leaderboard (e.g. only sort_type=1) when the activity has more than one.

Only ask the user to pick one when there are clearly too many to fit (≥ 3 leaderboards on a single competition). With 1–2 leaderboards, always show all by default. </MUST>

tabConfigs[].rankFieldConfig[] fields:

• title — display name (e.g. PnL%, PnL)
• key — internal sort field (e.g. pnl, realizedProfit)
• sortValueMap.descend — the numeric value to pass as --sort-type

Per-leaderboard fetch:

onchainos competition rank --activity-id <id> [--wallet <addr>] --sort-type <descend> --limit 20

Display rules: for each leaderboard render a separate section labeled by its title. Each section shows top N entries: rank, nickname (masked), score (userTotal formatted by format field), estimated reward.

Example response (activity with two leaderboards):

PnL% leaderboard — pool 200 DJT

Rank 1, Agentic....sMWP, PnL% +0.17%, estimated reward 100 DJT

Rank 2, Agentic....gweD, PnL% +0.03%, estimated reward 20 DJT

PnL leaderboard — pool 200 DJT

Rank 1, Agentic....sMWP, PnL $0.1885, estimated reward 100 DJT

Rank 2, Agentic....gweD, PnL $0.0006, estimated reward 20 DJT

After the leaderboards, append a "Your rank" section using the CASE 1 / 2 / 3 templates from the next section, since you already have all the data.

Check user's own rank (across ALL leaderboards)

A user can simultaneously appear on multiple leaderboards (e.g. PnL% AND PnL). When the user asks "what's my rank?", you MUST query every leaderboard the activity exposes, then render one of the three fixed templates below.

Required flow:

1. Call competition_detail → enumerate tabConfigs[].rankFieldConfig[].sortValueMap.descend to get the full set of sort_type values for this activity.
2. For EACH sort_type, call competition_rank --sort-type <descend> and capture myRankInfo plus the leaderboard's threshold (lowest userTotal in allRankInfos).
3. Classify the result:

• CASE 1 — user has currentRank > 0 on every leaderboard
• CASE 2 — user has currentRank > 0 on at least one but not all
• CASE 3 — user has no currentRank > 0 on any leaderboard

1. Output the matching fixed template, rendered in the user's language (English canonical below; localize for Chinese / other-language users).

<MUST> Output exactly the matching template structure below — never paraphrase the data fields, never collapse the two-leaderboard sections into one. Localize the natural-language strings to the user's language; keep placeholders, numeric values, and units verbatim. </MUST>

CASE 1 — ranked on both PnL and PnL%

Template:

Realized PnL ranking:
You are currently ranked #{pnlRank}, estimated reward {pnlReward} {rewardUnit}!
 
Realized ROI ranking:
You are currently ranked #{roiRank}, estimated reward {roiReward} {rewardUnit}!
 
| Leaderboard | My rank | Estimated reward |
|-------------|---------|------------------|
| Realized PnL | #{pnlRank} | {pnlReward} {rewardUnit} |
| Realized ROI | #{roiRank} | {roiReward} {rewardUnit} |
 
Your total estimated reward across both rankings: {totalReward} {rewardUnit} (sum of the two)

CASE 2 — ranked on one leaderboard, off the other

There are two symmetric sub-cases. The structure is identical: the ranked leaderboard goes first ("ranked #N, estimated reward X"), then the unranked one ("not on the leaderboard, current value Y, threshold Z"). Each sub-case has its own pinned template — do NOT improvise the unranked-section unit (% for PnL%, currency $ for PnL).

CASE 2-A — on PnL, off PnL% (currentRank for sort_type=7 > 0; sort_type=1 == 0)

Template:

Realized PnL ranking:
You are currently ranked #{pnlRank}, estimated reward {pnlReward} {rewardUnit}!
 
Realized ROI ranking:
Not on the leaderboard yet. Your current realized ROI is {currentRoi}%. You need at least {minRoi}% (the current leaderboard minimum) to qualify.

CASE 2-B — on PnL%, off PnL (currentRank for sort_type=1 > 0; sort_type=7 == 0)

Template:

Realized ROI ranking:
You are currently ranked #{roiRank}, estimated reward {roiReward} {rewardUnit}!
 
Realized PnL ranking:
Not on the leaderboard yet. Your current realized PnL is ${currentPnl}. You need at least ${minPnl} (the current leaderboard minimum) to qualify.

Section ordering rule: the leaderboard the user IS ranked on ALWAYS goes first. Don't put the "Not on the leaderboard" section before the ranked one.

Unit rule: PnL% uses % suffix (no currency symbol); PnL uses $ prefix (or the appropriate currency unit). Do NOT mix them up — the user's threshold for PnL is a dollar amount, not a percentage.

CASE 3 — off both leaderboards

Template:

Your address is not on any leaderboard. Your current realized PnL is ${currentPnl}, realized ROI {currentRoi}%.
The current minimum to qualify: realized PnL ${minPnl}, realized ROI {minRoi}%.

Field-mapping rules

• {pnlRank} ← myRankInfo.currentRank of the PnL leaderboard (sort_type 7)
• {pnlReward} ← myRankInfo.expectedRewards of the PnL leaderboard
• {roiRank} ← myRankInfo.currentRank of the PnL% leaderboard (sort_type 1)
• {roiReward} ← myRankInfo.expectedRewards of the PnL% leaderboard
• {rewardUnit} ← myRankInfo.rewardUnit (e.g. DJT); per-leaderboard if they ever differ
• {totalReward} ← pnlReward + roiReward (numeric sum, same unit)
• {currentRoi} ← user's PnL% score from myRankInfo.userTotal of the PnL% board (or 0 if backend returned null)
• {currentPnl} ← user's PnL score from myRankInfo.userTotal of the PnL board
• {minRoi} ← lowest qualifying PnL% — last entry's userTotal in the PnL% board's allRankInfos[]
• {minPnl} ← lowest qualifying PnL — last entry's userTotal in the PnL board's allRankInfos[]

If the activity exposes leaderboards beyond PnL/PnL% (future expansion via tabConfigs[]), extend the same template pattern: one section per leaderboard, summary table aggregates all, total reward sums all expectedRewards.

format: 1=number, 2=percentage, 3=token amount with unit.

Step 6 — Claim Reward

Check status first via competition_user_status:

Pre-claim guard (rewardStatus=4 / Pending draw)

<MUST> When the user explicitly requests to claim a reward (any "claim my reward" / "claim X" intent in any language) for an activity whose rewardStatus is 4 (Pending draw), do NOT call competition_claim. Render the Pending-draw canonical template (see Step 5 above, with {activityName} substituted) instead.

This applies whether the user explicitly named the activity or you inferred it from prior status output. Calling claim on a rewardStatus=4 activity would either be rejected by the backend or, worse, returns a confusing technical error. The canonical template is the only correct user-facing response. </MUST>

Atomic claim (the only correct path)

Both the MCP tool competition_claim and the CLI onchainos competition claim now do the same atomic flow: pre-check rewardStatus, fetch calldata, sign each entry with the TEE session, broadcast on-chain, return txHash array. The CLI no longer returns raw unsigned calldata — the only externally visible behavior is the final result.

MCP (preferred when running inside Claude Code / any AI environment):

competition_claim(activity_name="...")  →  { rewardAmount, rewardUnit, succeeded[], failed[] }

CLI (terminal use, or AI shelling out via Bash):

onchainos competition claim --activity-id <id> --evm-wallet <evm_addr> --sol-wallet <sol_addr>
# → returns the same { rewardAmount, rewardUnit, succeeded[], failed[] } shape

Result shape (both paths):

{
  "rewardAmount": "460",
  "rewardUnit": "PYBOBO",
  "totalEntries": 1,
  "succeeded": [{"contractAddress": "...", "chain": "501", "txHash": "...", "orderId": "..."}],
  "failed": [],
  "needContact": false,
  "activityId": "107",
  "accountId": "5747d742-...",
  "joinedAddress": "0x8e3f..."
}

How to report to the user:

• All succeeded (failed: []): "Claimed {rewardAmount} {rewardUnit}, tx hash: {txHash}"
• Partial success (some failed): list each succeeded txHash, then list the failed entries with their error, then append the fixed failure-suggestion block (template below). Do NOT re-run claim blindly — succeeded entries already landed; another call will hit the "reward already claimed" guard.
• All failed: the tool returns an error, not this shape — surface the error message verbatim, then append the fixed failure-suggestion block.
• If `needContact: true` in the response (user is a top-tier winner who has NOT yet submitted contact info): after the success line above, also render the Contact-collection prompt below — invite (do NOT force) the user to share one contact method. See #### Contact collection (top-tier winners only) further down for the prompt template, parsing rules, and follow-up.

The flow blocks before signing if rewardStatus is 0 (not eligible), 2 (already claimed), 3 (expired), or 4 (winners not announced yet). The error message is plain text — relay it to the user. Skip the failure-suggestion block in these pre-check rejections (they are semantic, not transient — telling the user to "check Gas / try later" is misleading).

Fixed failure-suggestion block

<MUST> For runtime failures (signing/broadcast/simulation errors, network errors, unknown errors), append this block after the error description. Translate to the user's language while preserving the heading + 3 bullet items in this order. Do NOT add or remove items. </MUST>

Template:

Suggestions:
- The claim process requires Gas. Please make sure your Gas is sufficient.
- Try again later — this may be a transient network issue.
- If it fails repeatedly, please contact customer support.

<NEVER>

• ❌ Do NOT show this block on pre-check rejections (rewardStatus=0/2/3/4) — the issue is not Gas / not transient.
• ❌ Do NOT show this block on code=11002 (not won) or code=11008 (claim expired/already claimed) — same reason.

</NEVER>

<NEVER>

• ❌ Do NOT chain gateway_broadcast after a claim call — the on-chain submission already happened inside the tool.
• ❌ Do NOT manually construct, encode, or sign a transaction (no Python base58 encoding, no manual hex assembly). The TEE-managed wallet key is the only valid signer.
• ❌ Do NOT inspect the result for an empty base58CallData and conclude the CLI cannot sign a Solana claim — that field is empirically empty for Solana; the CLI/MCP code internally falls back to encoding tx.data byte array via base58 and proceeds. Just trust the succeeded[] and failed[] arrays.
• ❌ Do NOT split into a two-step "fetch calldata then wallet contract-call" flow — that mode no longer exists; the claim command is atomic.

</NEVER>

On claim error (code 11002 `not eligible for reward`): "You did not win a reward and cannot claim." On any other error: "Operation failed. Please contact customer support."

Contact collection (top-tier winners only)

<MUST> Run this sub-flow if and only if the competition_claim response contains needContact: true. Do NOT run it when needContact: false or the field is missing. Do NOT ask for a contact pro-actively in any other claim path. </MUST>

Step 6a — After the claim-success line, append this prompt (English canonical; translate the natural-language strings to the user's conversation language; keep the 4 numbered options in this exact order and the literal labels Telegram / WeChat / Email / Twitter (X) as-is — these are product-canonical, do not paraphrase):

Congratulations on your standout performance in this competition! As a thank-you, we have a custom merchandise pack reserved for top winners. Please share ONE of the following contact methods so we can reach out about delivery — sharing is optional:
 
1. Telegram
2. WeChat
3. Email
4. Twitter (X)

Step 6b — When the user replies with a contact method, parse out contactType and contactValue from their message:

contactType MUST be one of the four exact case-sensitive strings (Telegram, WeChat, Email, Twitter) — the backend rejects anything else. If the user's message is ambiguous (e.g. just @username with no platform), ask once which platform they meant; do NOT guess.

Step 6c — Call `competition_submit_contact`:

competition_submit_contact(
  activity_name="<same activity name used in competition_claim>",
  contact_type="Telegram" | "WeChat" | "Email" | "Twitter",
  contact_value="<the parsed value, max 256 chars>"
)

CLI equivalent:

onchainos competition submit-contact --activity-id <id> --contact-type Telegram --contact-value "@testemma"

Step 6d — On `submitted: true` response, render this confirmation (English canonical; translate to the user's language; do NOT echo the contact value back; do NOT show any internal id):

Got it. Thanks for sharing! We will reach out shortly — please keep an eye on your messages.

On submit_contact error, surface the message verbatim with a short hint:

• If the backend returns a validation error on contactType, re-prompt the user with the 4 options.
• If not registered for activity — this should never happen post-claim; flag as a backend anomaly and tell the user to retry later.
• Other errors: "Failed to record your contact, please try again later or contact customer support."

<NEVER>

• ❌ Do NOT ask for contact on regular (non-top-tier) claims — needContact: false means do nothing.
• ❌ Do NOT echo the contact value back to the user in the confirmation message — they already see it in the conversation; repeating it can feel intrusive.
• ❌ Do NOT push / pressure the user if they decline to share. Acknowledge politely and move on.
• ❌ Do NOT prompt for multiple contacts — one is enough. Stop after the first valid submission.
• ❌ Do NOT call competition_submit_contact proactively without the user explicitly providing a contact value in this conversation turn.
• ❌ Do NOT expose activityId / accountId / the parsed contactType enum string to the user in any rendered message.

</NEVER>

Additional Flows

Query Registered Wallet

When user asks "show my registered address" or similar:

1. Call competition_user_status (MCP) — accountId is loaded from the active wallet session; no wallet args needed. CLI equivalent: onchainos competition user-status (omit --activity-id to query all activities).
2. Find entries where joinStatus=1
3. For each matched entry, present: competition name (activityName) + chain (chainName) + masked address (first4...last4). Use chain to determine which address was used (EVM or SOL).

If multiple entries match, list all of them.

Example (single):

Your Account 1 is registered for XXX Trading Competition. Registered address: Solana address DeEV...Fbx.

Example (multiple):

Your Account 1 is registered for the following trading competitions:

- XXX Trading Competition (Solana): DeEV...Fbx

- YYY Trading Competition (XLayer): 0x1234...abcd

If no entry has joinStatus=1:

You are not currently registered for any trading competition.

Wallet Export Guard

When the user requests to export the Agentic Wallet:

1. Call competition_user_status (MCP) — uses accountId from active session. CLI equivalent: onchainos competition user-status.
2. If any joinStatus=1:

> Your wallet is registered for an Agentic Wallet trading competition. Exporting the wallet will forfeit your eligibility for this competition. Please confirm whether you want to proceed with the export.

1. Only proceed with export if the user explicitly confirms.

Status Codes

--status filter parameter (input only)

Response field values

Error Handling