DJ Prompts

Conversational prompts that combine the agent’s tools with its musical knowledge. No special setup needed beyond having your caches populated — just paste a prompt and go.

Gig Prep

Context-aware set preparation. Instead of generic “build a set,” tell the agent about the gig and let it translate context into constraints.

Paste into your agent:

I have a gig coming up. Here's the context:

- Venue/event: [name, type — club, festival, bar, warehouse]
- Time slot: [e.g., 1-3am, opening, closing, sunrise]
- Duration: [e.g., 2 hours]
- Who's playing before/after me: [if known]
- Vibe I'm going for: [in your own words]
- Anything I definitely want to play: [optional track names]

Help me prepare.

What the agent should do:

Translate the context into musical constraints:
- Time slot → energy arc shape (opening = slow build, peak = sustained high, closing = controlled descent, sunrise = floaty/euphoric)
- “After an industrial DJ” → don’t go harder, build tension differently
- “Bar, 50 people” → lower energy ceiling, more variety, deeper cuts
Use search_tracks to find tracks matching the derived constraints (BPM range, genre, energy range)
Use query_transition_candidates to identify tracks that work well together within the pool
Present a curated pool (not a fixed sequence) — the DJ picks the order live
Optionally run build_set for a suggested sequence within the pool

The value is in the translation from natural language context to musical parameters — something a GUI tool can’t do.

Collection Gap Analysis

Understand your library’s shape — and its blind spots.

Paste into your agent:

Analyze my collection for gaps and imbalances. I want to understand:
- Where am I deep vs. thin?
- What keys/tempos/genres am I missing?
- Am I overindexing on anything?
- What limits my flexibility as a DJ?

What the agent should do:

Use read_library for overall stats (genre distribution, key distribution, track count)
Use search_tracks with various filters to map coverage:
- BPM buckets (< 120, 120-125, 125-130, 130-135, 135+) — where are you thin?
- Key coverage across the Camelot wheel — any dead ends where you have 1 track and nothing compatible?
- Genre coverage — are you a techno DJ with 3 house tracks, or evenly spread?
- Energy distribution — all bangers and no warmup material?
- Rating distribution — are you only rating the tracks you already love?
Identify specific gaps:
- Keys with < 3 tracks (harmonic dead ends)
- BPM ranges with < 5 tracks (can’t sustain a set there)
- Labels you buy from heavily — use lookup_discogs to check for recent releases you might be missing
Present findings as actionable insights, not just charts

Dig Session Partner

A knowledgeable collaborator for music discovery, using your collection as context.

Paste into your agent:

I'm digging for new music today. Here's what I'm looking for:
[describe what you want — genre, mood, tempo, reference tracks, or just "surprise me"]

Use my collection to understand my taste and suggest directions.

What the agent should do:

Analyze the user’s collection for patterns:
- Most common labels → suggest sibling/adjacent labels
- Artist clusters → suggest related artists via Discogs credits (remixers, collaborators, label-mates)
- Genre/era concentrations → suggest unexplored adjacent territories
Use lookup_discogs and lookup_beatport to research suggestions
Cross-reference against what the user already owns — don’t suggest what they have
Notice purchasing patterns:
- “You’ve bought 6 tracks from this label but nothing from their last 2 years — worth checking”
- “3 of your recent purchases are from artists who all appear on [compilation] — the other artists on it might interest you”
- “Your collection is 80% post-2020 releases — want some classic material for contrast?”
Suggest specific releases or tracks to check out, with reasoning

Post-Gig Debrief

Analyze what you actually played and learn from it.

Paste into your agent:

Debrief my last gig. Analyze the session from [date] and help me understand:
- What was my energy arc?
- What's in heavy rotation vs. fresh?
- How did my actual set compare to what I prepped?

What the agent should do:

Use get_sessions to find the session, get_session_tracks for the tracklist
Resolve full track data and analyze the arc:
- Energy curve — where were the peaks and valleys? Was the peak too early/late?
- BPM trajectory — did tempo drift up, stay flat, or follow a deliberate arc?
- Harmonic movement — smooth Camelot steps or big jumps? Were the jumps intentional?
- Genre flow — did you stay in one lane or move between styles?
Cross-reference with play history:
- Tracks appearing in 3+ recent sessions → heavy rotation (signature tracks or crutches?)
- Tracks played for the first time → how did they fit?
- Tracks from the prepped pool that weren’t used → why not?
Surface actionable observations:
- “Your first hour was all 6A/7A — harmonically safe but maybe too static”
- “You jumped from energy 4 to 8 at track 12 — was that intentional or did you panic?”
- “You played ‘Black Sun’ at your last 4 sessions — maybe rest it?”

Harmonic Journey Planning

Plan extended harmonic arcs across a set — beyond pairwise key compatibility.

Paste into your agent:

Plan a harmonic journey for my next set.
- Starting key: [e.g., 6A, or "wherever my opener is"]
- Style: [rising tension / major-minor shift / full wheel rotation / stay close]
- Duration: [number of tracks or minutes]
- Pool: [playlist name, genre filter, or "my whole library"]

What the agent should do:

Map the user’s available tracks by Camelot position using search_tracks
Identify the harmonic landscape — where are the dense clusters? Where are the gaps?
Plan a key sequence based on the requested style:
- Rising tension: move clockwise around the wheel (6A → 7A → 8A → …), each step creates harmonic lift
- Major-minor shift: move between inner and outer ring (6A → 6B → 7B → 7A) for emotional contrast
- Full rotation: traverse the full wheel (12 positions on the outer or inner ring) and return to start
- Stay close: never move more than 1 step, maximize harmonic smoothness
For each position in the journey, suggest specific tracks that fit, scored by query_transition_candidates
Flag problems:
- “You have nothing in 12A — this journey can’t pass through there”
- “The jump from 9A to 11A skips a position — you’ll need a bridge track or an intentional energy shift”
- “Your 3B tracks are all above 135 BPM but your 2A tracks are all below 125 — the key transition will also be a big tempo jump”

Practice Session Design

Structured practice instead of aimless browsing.

Paste into your agent:

Design a practice session for me.
Focus: [e.g., "learn my new tracks", "practice difficult transitions",
        "explore a key I never use", "get comfortable below 124 BPM"]
Duration: [e.g., 30 min, 1 hour]

What the agent should do:

Based on the focus area, select appropriate tracks:
- New tracks: find recently added tracks (by date or specific batch), pair them with familiar tracks for context
- Difficult transitions: find pairs with challenging but rewarding compatibility — key jumps that work with the right timing, BPM gaps that need manual adjustment
- Weak key areas: find tracks in keys the DJ rarely uses (from gap analysis), pair with comfortable keys to practice moving in and out
- Tempo range: find tracks in the target BPM range, pair in sets of 3-4 for extended practice
Use score_transition to find pairs that are:
- Achievable but not trivial (score 60-80 range — too easy doesn’t build skill, too hard frustrates)
- Varied in challenge type (some harmonic, some tempo, some energy)
Present as ordered exercises:
- “Mix 1: Track A → Track B (compatible keys, practice the 4 BPM gap)”
- “Mix 2: Track C → Track D (key jump from 6A to 8A — try mixing during the breakdown)”
- “Mix 3: Track E → Track F → Track G (3-track chain, maintain energy while shifting keys)”
After practice, ask what worked and what didn’t — refine for next session