Agent skill
tarot
Draw tarot cards as a thinking tool for technical decisions, blockers, and creative reframing. Use this skill when someone says "pull a card", "draw a card", "tarot", "tarot reading", "what do the cards say", or "card pull". Also use when someone asks for a symbolic or lateral reframe on a specific decision they're stuck on — like choosing between auth approaches, database architectures, or migration strategies — and they use language like "give me a different angle", "I need a lens on this", or "reframe this". Use for opening draws before work sessions and closing cards after milestones. Agents in orchestrated flows (idea-girlie researchers, team leads, planners) can invoke this as a composable thinking tool whenever a lateral perspective shift would help unstick a problem.
Install this agent skill to your Project
npx add-skill https://github.com/Temple-of-Silicon/esoterica/tree/main/plugins/tarot/skills/tarot
SKILL.md
Tarot
Tarot is a lens, not a prediction engine. You draw cards, and the archetypal imagery gives you a frame to see your situation differently. That's all. That's enough.
Claude already knows tarot deeply — every card, every spread, every tradition. This skill doesn't teach card meanings. It teaches how to read well and makes tarot available as a composable tool for agents and humans alike.
Drawing Cards
Use shell randomness for genuine draws. Don't pick cards yourself.
shuf -i 0-21 -n 1 # single Major Arcana
shuf -i 0-21 -n 3 # three Major Arcana
shuf -i 0-77 -n 1 # single from full deck
shuf -i 0-77 -n 3 # three from full deck
Card mapping: 0-21 = Major Arcana (The Fool through The World), 22-35 = Wands, 36-49 = Cups, 50-63 = Swords, 64-77 = Pentacles. Each suit: Ace(0), Two(1)...Ten(9), Page(10), Knight(11), Queen(12), King(13) offset from suit start.
After drawing, interpret from your own knowledge.
Reading Approach
The one rule: Interpret FOR them, not AT them. Echo their situation back through the card — don't just describe the card and leave them to make connections.
Good: "You mentioned the auth refactor keeps stalling — The Tower suggests this isn't procrastination, it's the structure resisting because it knows it needs to fully collapse."
Bad: "The Tower represents sudden upheaval and the destruction of false structures."
Calibrate depth to investment:
- Quick draw (no context): ~150-200 words
- Standard (some context): ~250-300 words
- Deep (rich context): ~350-400 words
Multi-card spreads: Weave cards into ONE story, not separate sections. No ## Card 1, ## Card 2 headers. Positions flow into prose. Name card relationships explicitly — where they conflict, reinforce, or transform each other.
Always end with a specific reflective question tied to their context and the card(s) drawn. Never generic ("what do you think?"). Always specific ("given The Tower's collapse and The Hermit's counsel, what's the one thing you've been avoiding looking at directly?").
Include shadow when relevant. Cards have shadow sides. If it fits their situation, name it gently.
Spreads
- Single card — One focused interpretation. The default.
- Three-card (Situation/Action/Outcome) — The workhorse. What's present, what to do, where it leads.
- Agent-suggested — Generate position names specific to the querent's question. More illuminating than generic positions. Example for "my team is burned out": What's Draining / What's Sustaining / What Would Renew.
- Custom — The querent (human or agent) names 1-5 positions. Honor their exact language.
When a human invokes directly, ask what they want. When an agent invokes as part of a flow, the agent chooses.
Physical Mode
Some humans read with physical decks. When physical_mode=true is in config (.tarot or ~/.claude/tarot/config), offer the option to draw from a physical deck.
Flow: ask them to shuffle and draw, accept card names or numbers generously ("tower", "The Tower", "16" all work), confirm before interpreting. If config doesn't enable physical mode, default to digital — don't ask, just draw.
Agent Composability
This skill is designed to be invoked by other agents and skills, not just humans. Tarot is a lens that adds value in many contexts:
Opening ceremony — Pull a card to set intention before a work session. Brief, framing, not predictive.
Closing reflection — After a session or milestone. Interpret through what was accomplished and what's ahead.
Reframe — When stuck or in a loop. The card provides a lateral perspective shift. This is where tarot is most useful as a thinking tool.
Problem-solving — Three-card spread on a specific challenge, with agent-suggested positions custom to the problem.
Ritual integration — Skills like micro-ritual, sacred-blessing, or correspondence can invoke tarot as one element of a composed experience.
When invoked by another agent:
- The calling agent provides context and can specify spread, card count, and deck
- Defaults: single card, major arcana only
- Return the reading as text — the calling agent decides how to present it
Config
Config lives in .tarot (project) or ~/.claude/tarot/config (global). Project overrides global.
physical_mode=false # enables physical deck option
deck=major # or full
All optional. Defaults: digital mode, major arcana only.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
generate-image
Generate images using Replicate's Nano Banana Pro model with Esoterica's eco-futurist divine feminine aesthetic. Use this skill when someone says "generate an image", "create a visual", "hero image", "make me an image", or wants visuals in the 1980s-meets-2180s Kodak film aesthetic. Also use when creating social assets, exploring visual variations, or producing branded imagery for Esoterica content. Agents producing content or building ritual experiences can invoke this to generate accompanying visuals.
correspondence
Find the symbolic vocabulary for any intention — herbs, colors, stones, times, directions, and elements that align with what you're working toward. Use this skill when someone says "what corresponds to", "correspondences for", "what herbs for", "what color for", "what stone for", "symbolic associations", or asks about magical ingredients for an intention. Also use when someone is building an altar, choosing outfit colors with intention, or encoding meaning into a space or action. Agents composing rituals, blessings, or creative work can invoke this to ground symbolic choices in traditional correspondence.
incantation
Generate a single-sentence power phrase for a specific moment or intention — not an affirmation (passive) but an incantation (active), words meant to be spoken aloud or carried into action. Use this skill when someone says "give me an incantation", "I need words for", "power phrase", "words to carry into", or "something to say before". Also use when someone is facing a specific moment — a meeting, a confrontation, a creation — and needs words that declare rather than describe. Agents composing rituals or ceremonial flows can invoke this to add a spoken-word element to a composed experience.
sacred-blessing
Create personalized sacred blessings for tasks, transitions, or life moments using inclusive feminine spiritual language. Use this skill when someone says "bless this", "I need a blessing", "sacred blessing", "give me a blessing for", or "bless my [thing]". Also use when someone is about to begin something meaningful — a medical procedure, a creative project, a hard conversation — and wants words of consecration. Agents composing rituals or ceremonial flows can invoke this as one element alongside tarot, correspondence, or micro-ritual.
micro-ritual
Design simple, embodied micro-rituals for life moments using ordinary objects at hand. Use this skill when someone says "create a ritual", "I need a ritual for", "ritual for this moment", "mark this transition", or "how do I make this feel sacred". Also use when someone describes a transition, threshold, or moment they want to honor with intention — starting a new job, closing a chapter, facing something hard — and wants a grounded, practical way to mark it. Agents in orchestrated flows can invoke this to compose rituals that weave in other elements like tarot draws, correspondence lookups, blessings, or incantations.
verl-rl-training
Provides guidance for training LLMs with reinforcement learning using verl (Volcano Engine RL). Use when implementing RLHF, GRPO, PPO, or other RL algorithms for LLM post-training at scale with flexible infrastructure backends.
Didn't find tool you were looking for?