The THIRI Chord Intelligence server is a deterministic music theory engine providing the following capabilities:
Analyze chords (analyze_chord): Parse any chord symbol (e.g. Dm7b5, G7#11) into its root, quality, intervals, and extensions. When a key is provided, returns the Roman numeral, scale degree, harmonic function (e.g. secondary dominants, modal interchange), and whether the chord is diatonic.
Resolve chords to notes (resolve_chord): Convert a chord symbol into enharmonically correct spelled note names, frequencies in Hz, MIDI numbers, and recommended improvisation scales.
Generate instrument-ready voicings (generate_voicing): Produce voicings in multiple styles — rootless, shell, triad, pad, guide-tones, drop-2, drop-3, Bill Evans, and more. Supports explicit tension/color preferences (b9, #11, b13, etc.), octave and density control, and voice-leading scoring by passing in the previous voicing.
Reharmonize progressions (reharmonize): Apply jazz reharmonization techniques including tritone substitution, ii-V insertion, modal interchange, diminished passing chords, secondary dominants, chain of dominants, Coltrane changes, and backdoor dominants — or use auto to get every applicable technique at once, each with an explanation.
Conduct a band (conduct_band): Generate musical lanes and MIDI data from natural-language band direction descriptions, suitable for further audio rendering.
Provides correct chord charts and music theory analysis for outputs from Suno or similar generative music platforms, enabling reliable chord labeling and reharmonization.
Give your AI real music theory. THIRI is the deterministic music theory MCP server + API for AI builders — it lets Claude, Cursor, or any MCP agent analyze chords, run roman-numeral analysis, generate voicings, and reharmonize progressions with answers that are computed, not guessed.
LLMs hallucinate music theory: wrong notes, fake roman numerals, voicings that don't voice-lead. THIRI is a deterministic engine (pitch-class-set theory over ℤ/12) behind a hosted API — so C7sus4 keeps its suspension, Caug spells C E G#, and "Coltrane changes on Dm7 G7 Cmaj7" returns Cmaj7 Ab7 Abmaj7 E7, every time.
Downstream of Suno / Udio or any generator? Wrap the output and get a correct chord chart your agent can trust. And unlike tonal.js or music21, THIRI is hosted and agent-native (no install, any language) — and it reharmonizes and voice-leads, not just looks chords up.
⭐ If this is useful, star the repo — it helps other musicians and agent builders find it.
👥 Join the First 55 AI Music Builders: Want elevated rate limits (300 req/min), direct founder support, and early access to upcoming tools? Join our developer community on Skool — Blues People AI.
Get a free key at build.thiri.ai/developers
In Claude: Settings → Connectors → Add custom connector → URL https://mcp.thiri.ai/mcp → paste your sk_live_ key
Ask Claude: "Reharmonize Dm7 G7 Cmaj7 with Coltrane changes."
That's it — no install, no config file. Builders: full install options (Claude Code, Desktop config, raw HTTP) are below.
Related MCP server: Music MCP
"Analyze Dm7b5 in C." →
iiø7, half-diminished, borrowed predominant, + scale options "What notes are in C7sus4?" →C F G Bb(the suspension survives) "Give me a rootless Cmaj7 voicing, then voice-lead into Dm7." → voicings + a voice-leading score "Reharmonize Dm7 G7 Cmaj7 with Coltrane changes." →Cmaj7 Ab7 Abmaj7 E7
Tool | What it does |
| Chord → root, quality, intervals, roman numeral & harmonic function (secondary dominants, modal-interchange labels) |
| Chord → spelled notes (enharmonically correct), frequencies, MIDI, scale recommendations |
| Instrument-ready voicings (rootless/bill_evans, shell, triad, pad, guide-tones, drop-2/3); pass |
| Progression reharmonization — 8 techniques: |
| Natural-language band conduct → lanes + MIDI (hosted MCP v0.3+) |
Runs on the v2 grid engine — correct sus chords, real triads, enharmonic spelling, all altered dominants — with request timeouts, quota reporting, and structured errors.
For hear-it agent loops (conduct → server-side render → WAV through your speakers), add a second local server alongside hosted theory tools:
{
"mcpServers": {
"thiri": {
"command": "npx",
"args": ["-y", "@bluesprincemedia/thiri-mcp"],
"env": { "THIRI_API_KEY": "sk_live_your_key" }
},
"thiri-conductor": {
"command": "npx",
"args": ["-y", "@bluesprincemedia/thiri-mcp", "thiri-conductor-mcp"],
"env": { "THIRI_API_KEY": "sk_live_your_key" }
},
"thiri-composition": {
"command": "npx",
"args": ["-y", "@bluesprincemedia/thiri-mcp", "thiri-composition-mcp"]
}
}
}Bin | Tools |
|
|
| Composition IR tools + |
Rendering runs server-side as of v0.5.0 — no Csound install needed. Proof: npm run test:conductor · live docs: build.thiri.ai/lab/conductor-mcp · agent recipes.
End-to-end persona for local vibe composition — skill, CLI, and Band dashboard panel:
Entry | Command / path |
Cursor skill | Copy |
CLI |
|
Dashboard |
|
Lab proof |
Dual MCP config above + mapConductResultToStudioModules after each conduct_band. Last CLI render writes ~/.thiri/conductor-last.json (local only, not committed).
Paste in order after dual MCP config above:
Analyze — "Analyze Dm7 G7 Cmaj7 in key C with analyze_chord; summarize roman numerals and tension."
Conduct — "conduct_band: warm Rhodes pad, walking bass, brush drums, 8 bars medium swing in C."
Render — "render_audio from the conduct result at tempo 120."
Critique — "play_audio; critique voice-leading and register balance; suggest one revision."
Full prompts: build.thiri.ai/lab/agent-recipes
Surface | Audio render |
| No — theory + |
Local | Yes — WAV rendered server-side ( |
Get a free key at build.thiri.ai/developers, then pick a path:
Claude Desktop / web / mobile — hosted (one-click custom connector, nothing to install):
Settings → Connectors → Add custom connector → URL https://mcp.thiri.ai/mcp → paste your sk_live_ key on the consent page. Same 5 tools, same key, same quota — no config file, no npx.
Claude Code (one line):
claude mcp add thiri --env THIRI_API_KEY=sk_live_your_key -- npx -y @bluesprincemedia/thiri-mcpClaude Desktop (claude_desktop_config.json):
{
"mcpServers": {
"thiri": {
"command": "npx",
"args": ["-y", "@bluesprincemedia/thiri-mcp"],
"env": { "THIRI_API_KEY": "sk_live_your_key" }
}
}
}The same engine is a plain REST API:
curl -X POST https://chords.thiri.ai/v2/analyze \
-H "Authorization: Bearer YOUR_KEY" -H "content-type: application/json" \
-d '{"chord":"Dm7b5","key":"C"}'Five endpoints: /v2/analyze, /v2/resolve, /v2/voicing, /v2/reharmonize, /v2/conduct. See openapi.yaml.
Variable | Default | Description |
| (none) | Bearer token ( |
|
| API base (override only for local dev) |
npm install && npm run build && npm startPolyForm Noncommercial 1.0.0 — © 2026 Blues Prince Media. Free for personal,
research, and noncommercial use; commercial use requires a license
(dennison@bluesprincemedia.com). See LICENSE. Versions published at or
before v0.5.0 remain under the MIT/PolyForm dual license they shipped with.
As of v0.5.0 the composition engine and Csound renderer run server-side behind the hosted API (
POST /v2/compose,POST /v2/render); their source no longer ships in this package.
Parse a chord symbol into its root, quality, intervals, extensions, and harmonic function. When a key is provided, returns the Roman numeral, scale degree, and whether the chord is diatonic.
| Name | Required | Description | Default |
|---|---|---|---|
| key | No | Key center for functional analysis (e.g. 'C', 'Bb') | |
| chord | Yes | Chord symbol (e.g. 'Dm7', 'Cmaj7/E', 'G7#11') |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations declare readOnlyHint=true, idempotentHint=true, and destructiveHint=false, which matches the parse-only behavior described. The description adds value by listing specific return components (root, quality, intervals, extensions, harmonic function) and conditional fields with key, beyond what annotations provide.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is two sentences, no wasted words, and front-loads the primary function. Every sentence adds value, making it efficient and easy to parse.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
No output schema exists, so the description bears full burden. It clearly explains return values for both with and without key scenarios, covering the main outputs. It could mention error handling or invalid input, but what's provided is sufficient for most use cases.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds meaning by explaining that providing a key returns Roman numeral, scale degree, and diatonic status, which is not in the schema. This clarifies the key parameter's effect on output, earning a 4.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description explicitly states it parses a chord symbol into root, quality, intervals, extensions, and harmonic function. It distinguishes itself from sibling tools like 'generate_voicing' (creates voicings) and 'reharmonize' (changes harmony), so purpose is clear and specific.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
No explicit when-to-use or when-not-to-use guidance is provided. Usage is implied by the tool name and description, but the description does not mention alternatives or exclusions, so guidance is insufficient.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Arrange a 4-piece band from a natural-language prompt. Returns conductor tempo/groove, 4 lanes of note events, lead sheet, and base64 MIDI. Browser/client synthesizes audio via Csound; use thiri-conductor-mcp locally to render WAV.
| Name | Required | Description | Default |
|---|---|---|---|
| prompt | Yes | Musical direction (e.g. '12-second neo-soul band in Eb, Rhodes + analog bass') | |
| durationSec | No | Optional target duration override |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Description adds value beyond annotations by specifying browser/client synthesis and local rendering path for WAV. Annotations already declare readOnlyHint=true, which is consistent as no external state is mutated.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences, no wasted words, front-loaded with core purpose. Every sentence adds value.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Describes outputs (tempo, lanes, lead sheet, MIDI) and synthesis context despite no output schema. For a complex creative tool, this is reasonably complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, baseline 3. Description adds meaningful example for prompt ('12-second neo-soul band in Eb...') and clarifies durationSec as optional override, enhancing understanding.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Specific verb 'Arrange a 4-piece band' with clear resource and outcome. Distinct from sibling chord-analysis tools.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
States output and local rendering requirement explicitly, providing context for use vs alternatives. No when-not-to-use guidance, but siblings are clearly different.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Generate instrument-ready voicings for a chord in a specified style. Pass previousNotes (or a previousVoicing object) from an earlier result to get a voiceLeadingScore for the transition. Styles: rootless, bill_evans, shell, triad, pad, guide-tones, guide-tone-1, guide-tone-2, both-guide-tones, drop-2, drop-3.
| Name | Required | Description | Default |
|---|---|---|---|
| chord | Yes | Chord symbol (e.g. 'Dm7') | |
| style | No | Voicing style (default: 'pad'). bill_evans is an alias of rootless. | |
| octave | No | Base octave (default: 3) | |
| density | No | Target voicing density / note count hint | |
| keyContext | No | Key context for diatonic color hooks (e.g. 'C major') | |
| previousNotes | No | Previous voicing notes for voice leading (e.g. ['E3', 'G3', 'Bb3', 'D4']) | |
| previousVoicing | No | Prior generate_voicing JSON result for strict guide-tone line tracking | |
| colorPreferences | No | Explicit color-tone overrides |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint=true, idempotentHint=true, and destructiveHint=false, so the safety profile is clear. The description adds behavioral detail (returns a voiceLeadingScore) without contradicting annotations.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two sentences: first defines the core purpose, second adds voice leading usage. No wasted words, front-loaded, and all information is relevant.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
With no output schema, the description could detail the return format (e.g., a JSON object with voicing notes and score). The current description mentions 'instrument-ready voicings' and 'voiceLeadingScore' but is vague on structure. Adequate but not complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema coverage is 100%, so baseline is 3. The description adds value by explaining the relationship between previousNotes and previousVoicing and returning a voiceLeadingScore, which goes beyond the individual parameter descriptions.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses a specific verb ('generate') and resource ('instrument-ready voicings') and lists all supported styles. It clearly distinguishes from sibling tools (analyze_chord, reharmonize, resolve_chord) which serve different purposes.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description explains how to use previousNotes or previousVoicing for voice leading, which provides clear context. It doesn't explicitly exclude scenarios or compare to siblings, but the context is sufficient for an AI to decide when to use this tool.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Reharmonize a chord progression using jazz techniques. Returns one alternative per applicable technique, each with the new progression, the changes, and a plain-English explanation. With technique 'auto' (default) it returns every technique that applies. Techniques: tritone_sub, ii_v_insertion, modal_interchange, diminished_passing, secondary_dominant, chain_of_dominants, coltrane_changes, backdoor. (modal_interchange and backdoor require a key.)
| Name | Required | Description | Default |
|---|---|---|---|
| key | No | Key center (e.g. 'C') — enables modal_interchange and backdoor | |
| technique | No | Reharmonization technique (default: 'auto' = all applicable) | |
| progression | Yes | Chord symbols, one per bar (e.g. ['Cmaj7', 'Dm7', 'G7', 'Cmaj7']) |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already provide readOnly and idempotent hints. Description adds behavioral details: returns one alternative per technique with progression, changes, explanation; 'auto' returns all applicable techniques. No contradictions.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
Two focused sentences plus a list and note. Front-loaded with core purpose. No extraneous words; every sentence adds value.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Adequately covers return format, technique behavior, and prerequisites. No output schema, but description explains outputs. Could briefly explain each technique, but not essential given jazz context.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
Schema covers all parameters with descriptions (100% coverage). Description adds value by explaining default technique behavior and key requirement for certain techniques, exceeding schema info.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
Clearly states 'Reharmonize a chord progression using jazz techniques' with specific verb and resource. Distinguishes from sibling tools (analyze_chord, generate_voicing, resolve_chord) by focusing on reharmonization. Lists techniques and return format.
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
Indicates when to use (reharmonization context) and notes conditions for some techniques (require key). Does not explicitly compare to alternatives, but sibling tool names imply distinct purposes.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Resolve a chord symbol to spelled note names, frequencies in Hz, MIDI numbers, and recommended improvisation scales.
| Name | Required | Description | Default |
|---|---|---|---|
| chord | Yes | Chord symbol (e.g. 'Cm7', 'F#dim7') |
Does the description disclose side effects, auth requirements, rate limits, or destructive behavior?
Annotations already declare readOnlyHint and idempotentHint, so the description's primary transparency contribution is listing the returned data types (note names, frequencies, MIDI numbers, scales). This adds useful behavioral context beyond what annotations provide.
Agents need to know what a tool does to the world before calling it. Descriptions should go beyond structured annotations to explain consequences.
Is the description appropriately sized, front-loaded, and free of redundancy?
The description is a single sentence that efficiently enumerates four distinct output categories. Every word contributes value, no repetition or fluff.
Shorter descriptions cost fewer tokens and are easier for agents to parse. Every sentence should earn its place.
Given the tool's complexity, does the description cover enough for an agent to succeed on first attempt?
Despite lacking an output schema, the description adequately lists the main return values. However, it could be more complete by noting error handling or format details. Still, for a simple lookup tool, it is reasonably complete.
Complex tools with many parameters or behaviors need more documentation. Simple tools need less. This dimension scales expectations accordingly.
Does the description clarify parameter syntax, constraints, interactions, or defaults beyond what the schema provides?
The single parameter 'chord' has 100% schema description coverage with examples. The description does not add extra meaning about the parameter syntax or format beyond the schema, so a baseline score of 3 is appropriate.
Input schemas describe structure but not intent. Descriptions should explain non-obvious parameter relationships and valid value ranges.
Does the description clearly state what the tool does and how it differs from similar tools?
The description uses the specific verb 'resolve' and identifies the resource as a chord symbol. It clearly lists the outputs (note names, frequencies, MIDI numbers, scales), which distinguishes it from sibling tools like analyze_chord (analysis) and generate_voicing (voicing).
Agents choose between tools based on descriptions. A clear purpose with a specific verb and resource helps agents select the right tool.
Does the description explain when to use this tool, when not to, or what alternatives exist?
The description implies this tool is for retrieving detailed chord data, but it does not explicitly state when to prefer it over siblings (e.g., analyze_chord, reharmonize). No usage context or exclusions are provided.
Agents often have multiple tools that could apply. Explicit usage guidance like "use X instead of Y when Z" prevents misuse.
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
v0.3.1conduct_bandv0.3.0analyze_chordgenerate_voicingreharmonizeresolve_chordScored across 5 tools
Each tool targets a distinct musical operation: resolving, analyzing, voicing, reharmonizing, and arranging. The only mild overlap is between resolve_chord and analyze_chord, since both accept chord symbols, but their output purposes are clearly separated by the descriptions.
Four of five tools follow a clear verb_noun pattern (resolve_chord, analyze_chord, generate_voicing, conduct_band). The exception is reharmonize, which is a single verb without an explicit object, but it is still recognizable and does not create confusion.
Five tools is well-scoped for a specialized music theory server. Each tool addresses a distinct and meaningful part of chord intelligence, from analysis and resolution to voicing, reharmonization, and full band arrangement.
The tool surface covers the core chord intelligence workflow: parse/analyze chords, resolve them to concrete musical data, generate voicings with voice-leading support, reharmonize progressions, and produce a full arrangement. There are no obvious dead ends or missing essential operations for the stated domain.
MCP server for progressive tool usage at any scale (see https://klavis.ai)
MCP server for Producer/Riffusion AI music generation
Generate AI music via the Lacuna Music API from MCP clients like Claude Desktop & Code.
AI Reasoning Cache & Consensus Layer with 11 MCP tools via Streamable HTTP.