What Shall We Build Next?

Describe

Describe your task

Refine

Refine the plan

SubAgents

Review all agents

Deploy

Deploy your agent

Let me break down the MEDIAAGENT subagent in detail:

A) SUBAGENT SUMMARY: 
Downloads a YouTube video and converts it to high-quality WAV format for audio processing, ensuring the audio is in the correct format for subsequent stem separation and beat analysis.

B) FINAL TASK OUTPUT:
A single high-quality WAV file (44.1kHz, 16-bit) containing the full audio from the YouTube video, saved to the server with a standardized filename format (e.g., "source_full_audio.wav")

C) SUBAGENT INPUT:
- YouTube video URL
- (Optional) Desired output filename

D) SUBAGENT TASK SUMMARY:
YouTube URL > #193 Download Entire YouTube Video As MP3 > #178 Convert 1-20 MP3s to wav > raw WAV file

E) DETAILED WORKFLOW:

1. Initial Input Processing:
   - Accept YouTube URL
   - Validate URL format
   - Generate standardized filename

2. Download Phase:
   - Use Skill #193 (Download Entire YouTube Video As MP3)
   - Input: YouTube URL
   - Output: MP3 URL on server

3. Conversion Phase:
   - Use Skill #178 (Convert 1-20 MP3s to wav)
   - Input: MP3 URL from previous step
   - Output: WAV URL on server
   - Format: 44.1kHz, 16-bit WAV

F) SILOS:
This subagent is simple enough that it doesn't require multiple silos - it operates in a linear fashion with each step feeding directly into the next.

NOTES:
1. This subagent is intentionally straightforward with minimal complexity, as its role is purely to prepare audio for the more complex processing that follows in other subagents.
2. The output quality is crucial as it affects all downstream processing.
3. No additional filtering or processing is done at this stage - that's handled by subsequent subagents.
4. Error handling should be implemented for both the download and conversion steps.

SubAgent #1 - Diagram

Expand Diagram

Let me break down SUBAGENT 2: ANALYSISAGENT in detail.

A) SUBAGENT SUMMARY: 
A specialized audio processing agent that analyzes an MP3 file to extract tempo information and separate it into vocal and instrumental stems, converting all outputs to WAV format for further processing.

B) FINAL TASK OUTPUT:
- 2x WAV files (vocal and instrumental stems)
- 1x JSON/text file containing:
  * Detected tempo (BPM)
  * Beat positions (in seconds)
  * Downbeat positions (in seconds)
All WAV files at 44.1kHz, 16-bit quality

C) SUBAGENT INPUT:
- 1x MP3 URL (from MEDIAAGENT)
- 3x timestamp positions (in seconds) for sample extraction

D) SUBAGENT TASK SUMMARY:
MP3 input > #180 Extract Beatpoints & Tempo > #197 Extract Instrumental & Vocal > #178 Convert MP3 stems to WAV > Final Output

E) SILOS:
The agent operates in three distinct silos:

SILO 1: TEMPO ANALYSIS
Input: MP3 URL
- Skill #180: Extract Beatpoints & Tempo of MP3
Output: Tempo data (BPM, beat positions, downbeat positions)

SILO 2: STEM SEPARATION
Input: MP3 URL
- Skill #197: Extract Instrumental & Vocal From MP3
Output: 2x MP3 URLs (vocal and instrumental)

SILO 3: FORMAT CONVERSION
Input: 2x MP3 URLs from Silo 2
- Skill #178: Convert MP3s to WAV
Output: 2x WAV URLs

The entire process flows:
1. Input MP3 is processed simultaneously through Silo 1 and Silo 2
2. Output from Silo 2 feeds into Silo 3
3. Final output combines tempo data from Silo 1 with WAV files from Silo 3

This structured approach ensures parallel processing where possible (tempo analysis can happen simultaneously with stem separation) and maintains clear data flow between components.

SubAgent #2 - Diagram

Expand Diagram

Let me break down SUBAGENT 3 (SAMPLEAGENT) in detail:

A) SUBAGENT SUMMARY: 
A specialized audio processing agent that takes vocal and instrumental stems and generates beat-aligned samples of varying lengths (1, 2, 4, and 8 beats) from three specified timestamp positions.

B) FINAL TASK OUTPUT:
Two collections of beat-aligned WAV audio samples (one for vocals, one for instrumentals):
- 3 timestamp positions × 4 sample lengths = 12 samples per stem type
- Each WAV file: 44.1kHz, 16-bit
- Named format: "[timestamp]sec_[stem-type]_[beat-length]beat.wav"
- Total output: 24 WAV files (12 vocal + 12 instrumental)

C) SUBAGENT INPUT:
- Two WAV files (vocal stem and instrumental stem)
- Three specific timestamps (in seconds)
- BPM/tempo data
- Beat position data (from previous agent)

E) SUBAGENT TASK SUMMARY:
For each stem type (vocal/instrumental):
1. Input WAV > #219 Cut WAV (at timestamp 1) > 4 samples (1,2,4,8 beats)
2. Input WAV > #219 Cut WAV (at timestamp 2) > 4 samples (1,2,4,8 beats)
3. Input WAV > #219 Cut WAV (at timestamp 3) > 4 samples (1,2,4,8 beats)

F) SILOS:
SILO 1: VOCAL STEM PROCESSING
- Input: vocal.wav + timestamps + tempo data
- Process: 3 rounds of #219 (one per timestamp)
- Output: 12 vocal samples (3 positions × 4 lengths)

SILO 2: INSTRUMENTAL STEM PROCESSING
- Input: instrumental.wav + timestamps + tempo data
- Process: 3 rounds of #219 (one per timestamp)
- Output: 12 instrumental samples (3 positions × 4 lengths)

Each silo runs independently but identically, using the same timestamps and beat calculations. The #219 skill is used repeatedly but with different parameters each time to generate the various beat-length samples. The agent automatically names and saves each file according to the specified naming convention.

SubAgent #3 - Diagram

Expand Flow

Let me break down SUBAGENT 4: PACKAGEAGENT in detail.

A) SUBAGENT SUMMARY: 
A packaging agent that organizes processed audio samples, generates metadata, and creates a structured zip archive of the final sample pack.

B) FINAL TASK OUTPUT:
A single .zip file containing:
- 24-48 .wav files (44.1kHz, 16-bit)
- Organized in folders (vocals/instrumentals)
- Named systematically (e.g., "130sec_vocal_4beat_1.wav")
- metadata.json file containing:
  - Original YouTube URL
  - Detected BPM
  - Sample timings
  - Number of samples
  - File organization structure

C) SUBAGENT INPUT:
- Collection of .wav files from SUBAGENT 3 (both vocal and instrumental)
- Original YouTube URL
- BPM/tempo data
- Original timestamp positions
- Beat lengths used for samples

E) SUBAGENT TASK SUMMARY:
1. Generate metadata (#223 Powerful LLM Prompt-to-Text Response)
   - Input: Collection of data points
   - Output: Structured JSON text

2. File organization and naming
   - Using built-in file organization function
   - Create folder structure
   - Rename files systematically

3. Create zip archive
   - Built-in zip function to package all files
   - Return final zip URL

F) SILOS:
SILO 1: METADATA GENERATION
- Purpose: Create structured metadata file
- Input: All technical data about samples
- Skills: #223
- Output: metadata.json file

SILO 2: FILE ORGANIZATION
- Purpose: Structure and rename files
- Input: All .wav files
- Skills: Built-in file organization function
- Output: Organized file structure with renamed files

SILO 3: ARCHIVE CREATION
- Purpose: Create final zip package
- Input: Organized files + metadata
- Skills: Built-in zip function
- Output: Final .zip file

The key insight here is that this subagent is primarily focused on organization and packaging, rather than audio processing. It needs to create clear, systematic organization of the files while preserving all technical metadata for future reference. The agent works in three distinct silos: metadata creation, file organization, and final packaging.