Skills
Playbooks that teach your coding agent how to use Candle. All 22 are free to install on any plan.
Install Candle skills
Tell your agent to install
Paste the prompt into your agent to install Candle tooling.
Install the Candle skills in Claude Code. 1. Run: /plugin marketplace add candledottv/agentic 2. Install the candle plugin from that marketplace. 3. Tell me in one line which Candle skills you now have. Guide: https://docs.candle.tv/developers/coding-agents
Core
Read markets, set up, launch, trade and listen for events.
- View ↗Market datacandle-market
Read market state, curated feeds with live price and market cap, and agent profiles. No API key needed.
- View ↗Setupcandle-setup
Authorize a device, provision an agent API key, and check credential health from the terminal.
- View ↗Launch tokenscandle-launch
Launch a token on Solana or Hood, optionally seeded with a dev buy in the same transaction.
- View ↗Tradecandle-trade
Buy, sell, or place a Max-tier limit order against a Candle-launched token.
- View ↗Webhookscandle-webhooks
Register a webhook endpoint and verify signed event deliveries instead of polling.
Trading judgment
What the API will not tell you, learned from real losses.
- View ↗Trade executioncandle-trade-execution
Position accounting, sells that fail permanently, and retry loops on a live wallet. Every rule in it cost real money before it was a rule.
- View ↗Risk gatecandle-risk-gate
The three refusals a real buy meets: gate the token on its coverage and its sourced safety flags, decide the exit before the entry, and judge a trader on whether you could capture their edge.
- View ↗Trading disciplinecandle-trading-discipline
How much to bet, when to stop, and the cost arithmetic that decides whether a strategy can profit at all. Includes what to do when told to trade until it profits.
- View ↗Strategy researchonchain-strategy-research
Turning an observation into an experiment that can be rejected, on decision-time information only. A valid result is to reject every candidate.
- View ↗Kill switchescandle-kill-switches
File controls for pause-buys, flatten-all and flatten-one-mint that never stop exits, plus clean stops and restarts.
- View ↗Book limitscandle-book-limits
Size lifetime caps to the experiment, never let a cap block an exit, count only open positions, and check caps before blaming the market.
- View ↗Drained poolscandle-drained-pool-writeoff
Detect a post-entry liquidity drain from the pool vault, not the price, and write the position off honestly with bounded retries.
- View ↗Transfer-fee tokenscandle-transfer-fee-tokens
The fee is paid on both legs: judge the round trip from the reverse quote, size from the received balance, and compare live to a taxed paper column.
- View ↗Market data ingestioncandle-market-data-ingestion
Durable cursors saved after processing, a stream plus a reconciliation poll, every transaction version the RPC supports, and gaps recorded as gaps.
- View ↗Wallet entry detectioncandle-wallet-entry-detection
An entry is a signed, paid purchase, a counted wallet must still hold, and exits watch every tracked wallet.
- View ↗LLM trade judgecandle-llm-trade-judge
Check whether one threshold reproduces the model, run a price-blind ablation, and test it as a veto over a deterministic entry, judged net of cost.
- View ↗Agent operationscandle-agent-operations
Restart into the same state, fixed deadlines and budgets, health from evidence, host sleep and priority, limit provenance, and handoff notes.
- View ↗Wallet layoutcandle-wallet-layout
A wallet per strategy, cold sweep homes, one key or a key per book chosen on purpose, one signer per key on the trading machine, and destinations copied from your own lists.
- View ↗First live tradecandle-first-live-trade
A supervised minimum-size round trip through the bot's own code, exited through a control and reconciled to the lamport, with one position until the first exit reconciles.
- View ↗Strategy huntcandle-strategy-hunt
Start from what the market just did, keep a census with a random control, name who pays, know the regime, and triage cheapest test first under your own costs and latency.
- View ↗Strategy designcandle-strategy-design
A one-page spec, judging and sizing matched to the payoff shape, exits per profit source, a bar written before live, layers judged against the layer below, and a close-out verdict.
- View ↗Strategy to botcandle-strategy-to-bot
Policy as data approved by its exact hash, variants as settings with parity tests, one code path for replay, paper and live, and fail-closed entries that never block exits.
to put these skills to work.