Trends CLI User Guide

Scope

• This skill is for end users operating the installed trends-skill-tool command.
• Installation channel is global npm only.

Trigger guardrails

• Primary Intent Trigger: Trigger whenever the user expresses intent related to:
  • Launching: "Launch a token", "create a memecoin", "launch on trends.fun", "deploy a coin".
  • Earnings: "Check my coin creator earnings", "how much have I earned?", "claim my rewards".
  • Launch History: "What tokens have I launched?", "my launch history", "show my created coins".
  • Portfolio: "What tokens do I hold?", "my holdings", "show my balance".
  • Transactions: "What transactions did I make?", "my transaction history", "recent trades".
  • Trading/Quotes: "Get a swap quote on trends.fun", "swap on trends.fun", "buy/sell on trends".
• Command-Based Trigger: Trigger for explicit commands: create, buy, sell, quote, reward, wallet, config, balance, holdings, created, transactions.
• Package-Based Trigger: Trigger when user mentions @trends-fun/trends-skill-tool, "trends command", or "trends tool".
• Lifecycle Trigger: Trigger when user asks to install, upgrade, or uninstall this CLI.
• Error Trigger: Trigger when user shares error messages containing "trends-skill" or related bonding curve failures.

Source-of-truth order

1. references/install-and-setup.md
2. references/command-recipes.md
3. references/error-playbook.md

Workflow

1. Identify the user intent:

• Install/setup
• Config/wallet
• Read-only query
• Trade flow
• Troubleshooting

1. Pick the matching recipe and provide runnable commands.
2. State expected observable outcomes.
3. Provide one concrete fallback path tied to known CLI behavior.
4. Recommend the next action only after the current step is verifiable.

Write Confirmation Gate (Trade Focused)

The following commands are write operations and require a confirmation gate by default:

• trends-skill-tool create
• trends-skill-tool buy
• trends-skill-tool sell
• trends-skill-tool reward claim

Non-gated commands (normal flow):

• trends-skill-tool balance
• trends-skill-tool holdings
• trends-skill-tool created
• trends-skill-tool transactions
• trends-skill-tool quote buy
• trends-skill-tool quote sell
• trends-skill-tool reward status
• trends-skill-tool wallet init
• trends-skill-tool wallet address
• trends-skill-tool config list|get|set|reset
• install/upgrade/uninstall commands

Use a staged flow for gated write commands:

• Phase 0 (parameter completion): collect missing parameters first. If anything is unresolved, ask focused follow-up questions and stop. Do not ask execution confirmation yet.
• Phase A (not confirmed): preflight analysis + fully resolved parameter echo. Do not provide final executable write steps.
• Phase B (confirmed): provide executable write steps.

For create, Phase 0 is mandatory unless the user already provided all fields or explicitly accepted defaults for missing fields.

Valid confirmation phrases include:

• confirm
• yes proceed
• 确认执行
• 直接执行

Bypass exception:

• If the user explicitly requests direct execution (direct write, 直接执行, 不要确认, skip confirmation), you may bypass the confirmation gate.
• Default bypass scope is current request only.
• If the user explicitly states session-level preference, apply bypass for the session.
• Even with bypass, still perform preflight disclosure for buy/sell quotes and reward claim status.
• Bypass does not skip Phase 0 parameter completion for missing required/optional create fields.

If confirmation is unclear:

• Ask for explicit confirmation and do not move to Phase B.

Write Intent Preflight Contracts

Before any gated write command, always provide a preflight summary.

For create preflight, always echo:

• name
• symbol
• url
• desc
• image source
• first-buy (initial buy amount at create time)
• dev-bps (token deployer share / 代币部署者分成) and split explanation

For create, treat users as beginners and explain fields in plain language:

• name: token name users will see.
• symbol: token ticker symbol users will see in trading views.
• desc: short reason/story for why the token exists.
• image: token icon source. If --image-path is missing, icon is auto-generated from symbol.
• url: related X profile or X tweet URL (x.com/twitter.com, profile or /status/...).
• first-buy: first buy amount (SOL) executed right after create.
• dev-bps: token deployer share (代币部署者分成, the user share). X creator share is 10000-dev-bps when url exists.

create parameter completion contract (must happen before confirmation):

• Do not ask confirm / yes proceed / 确认执行 until each field is resolved.
• A field is "resolved" only when:
  • user provided a value, or
  • user explicitly accepted the default/empty behavior.
• Resolve in this order:
  1. name and symbol (required, cannot be empty)
  2. desc (user text or explicit empty)
  3. image source (local --image-path or explicit auto-generate)
  4. url (valid X profile/tweet URL or explicit none)
  5. first-buy (explicit number or explicit 0)
  6. dev-bps (required decision only when url exists; otherwise mark ignored)
• If url exists and dev-bps not provided, ask whether to use default 9600.
• If url does not exist, explicitly state dev-bps is ignored and effective split is token deployer 100%, X 0%.
• If responding in Chinese during missing-parameter collection, use exact wording for URL prompt:
  • 请输入你想指向的费用受益人X profile 或推文链接（可留空）
• If responding in Chinese, label dev-bps as:
  • 代币部署者分成(dev-bps)
• During Phase 0, ask only missing-field questions plus default choices; do not ask execution confirmation.

create preflight rules:

• If --image-path is not provided, explicitly state image will be auto-generated from symbol.
• If first-buy is omitted or 0, explicitly state: create/pool initialization still happens, but no immediate first buy is executed.
• If first-buy is greater than 0, explicitly state: create and first buy are submitted atomically in one flow.
• If url is provided, show split between token deployer and X creator:
  • token deployer share = dev-bps (default 9600 if omitted)
  • X creator share = 10000 - dev-bps
• If url is not provided, explicitly state:
  • --dev-bps is ignored
  • effective split is token deployer 100%, X creator 0%

For buy/sell preflight, always:

• Run quote first (trends-skill-tool quote buy or trends-skill-tool quote sell)
• Echo route, input/output amount, slippage bps, and key quote fields
• State that buy/sell will run only after confirmation

For reward claim preflight, always:

• Run trends-skill-tool reward status first
• Echo accountExists, rewardLamports, rewardSol
• Only proceed to claim confirmation when accountExists=true and rewardLamports>0
• If not claimable, stop and provide retry condition

For holdings / created / transactions responses, always include identifiers:

• Always echo queried wallet address.
• For each token row, include token mint address (prefer keys in order: mintAddress, mint_addr, mint).
• Never summarize by name/symbol only; disambiguate duplicate symbols with mint address.
• Include next_cursor for created and transactions when present.
• If a desired field is unavailable in raw data, explicitly output N/A and include the raw key name you used.
• For transactions, include transaction identifier/signature/hash field when present.
• For numeric fields in holdings / created:
  • owner_context.balance is raw decimals=6
  • human-readable value = raw value / 1_000_000
• For numeric fields in transactions:
  • sol_amount is raw lamports (decimals=9), human-readable value = raw value / 1_000_000_000
  • token_amount is raw token base units (decimals=6), human-readable value = raw value / 1_000_000
• Always disclose conversions in the response and never present unconverted raw numbers as final user-facing amounts.

Secret Handling Policy (Hard Block)

Never request, read, print, or echo any of the following:

• Private keys
• Mnemonic/seed phrases
• secretKey arrays
• Raw keypair file contents

Never suggest commands that expose private key content, including:

• cat ~/.config/solana/id.json
• Any equivalent command that prints key material

If user provides private key material:

• Refuse to continue handling sensitive key content
• Instruct user to rotate/revoke compromised keys
• Continue only with address-level operations, such as trends-skill-tool wallet address

Mandatory response template

Use one of the following templates based on command type and confirmation state.

For read-only or non-gated tasks:

Situation

Summarize the user goal in one or two lines.

Commands to run

Provide copy-paste command blocks only with supported trends-skill-tool flags and subcommands.

Expected result

State what the user should see (key fields, success lines, or command behavior).

If it fails

Map to an exact error from references/error-playbook.md, then provide immediate fix commands.

Next step

Give the shortest validated next command or decision.

For gated write tasks before confirmation:

Situation

Summarize the user goal in one or two lines.

Preflight

Provide required preflight checks and parameter echo (including quote/status where required).

Pending write plan

Show what will be executed after confirmation. Do not provide final executable write steps yet.

Confirm needed

Ask for explicit confirmation (confirm / yes proceed / 确认执行 / 直接执行) only after Phase 0 parameter completion is finished.

If it fails

Map to exact known errors and immediate fixes.

Next step

State the shortest path to confirm and execute.

For gated write tasks after confirmation:

• Use the standard read-only template and provide executable steps.

Command policy

• Never invent commands or flags.
• Use only the supported command surface:
  • trends-skill-tool create
  • trends-skill-tool buy
  • trends-skill-tool sell
  • trends-skill-tool quote buy
  • trends-skill-tool quote sell
  • trends-skill-tool balance
  • trends-skill-tool holdings
  • trends-skill-tool created
  • trends-skill-tool transactions
  • trends-skill-tool reward status
  • trends-skill-tool reward claim
  • trends-skill-tool wallet init
  • trends-skill-tool wallet address
  • trends-skill-tool config list|get|set|reset
• Enforce mutually exclusive amount routes:
  • buy: exactly one of --in-sol or --out-token
  • sell: exactly one of --in-token or --out-sol
  • quote buy: exactly one of --in-sol or --out-token
  • quote sell: exactly one of --in-token or --out-sol
• Respect integer constraints:
  • Strict non-negative integers for options such as --slippage-bps, --dev-bps, --count, --compute-unit-limit, --compute-unit-price
  • created/transactions --count range is 1..25
  • holdings --count is >= 1
• Create-specific constraints:
  • Do not suggest --slippage-bps for trends-skill-tool create.
  • --first-buy means initial buy amount at create time:
    • omitted or 0: create only (no initial buy execution)
    • greater than 0: create + first buy submitted atomically
• Reward-specific constraints:
  • trends-skill-tool reward status is read-only and should be the first check before claim.
  • trends-skill-tool reward claim should be suggested only when reward account exists and reward is greater than 0.
  • If reward account is missing or reward is 0, explain that claim returns validation error and no transaction is sent.

Output policy

• Keep steps concise and operational.
• Prefer --json when the user requests parsing, automation, or script integration.
• Include concrete verification commands (trends-skill-tool --version, trends-skill-tool --help, targeted quote/query checks) when relevant.

Troubleshooting policy

• First, match exact known errors from references/error-playbook.md.
• Provide deterministic fix commands before broad diagnosis.
• If no exact match exists, run a minimal diagnosis set in this order:
  1. trends-skill-tool --version
  2. trends-skill-tool --help
  3. trends-skill-tool config list
  4. trends-skill-tool wallet address
  5. Re-run the failing command with explicit endpoints or --json as needed
• Do not provide speculative root causes without a validation step.

Reference routing

• For install, upgrade, uninstall, wallet bootstrap, config precedence: read references/install-and-setup.md.
• For normal task execution patterns: read references/command-recipes.md.
• For error-to-fix mappings and verification commands: read references/error-playbook.md.