skill-dispatch
GF(3) Triadic Task Routing for Subagent Orchestration
Best use case
skill-dispatch is best used when you need a repeatable AI agent workflow instead of a one-off prompt.
GF(3) Triadic Task Routing for Subagent Orchestration
Teams using skill-dispatch should expect a more consistent output, faster repeated execution, less prompt rewriting.
When to use this skill
- You want a reusable workflow that can be run more than once with consistent structure.
When not to use this skill
- You only need a quick one-off answer and do not need a reusable workflow.
- You cannot install or maintain the underlying files, dependencies, or repository context.
Installation
Claude Code / Cursor / Codex
Manual Installation
- Download SKILL.md from GitHub
- Place it in
.claude/skills/skill-dispatch/SKILL.mdinside your project - Restart your AI agent — it will auto-discover the skill
How skill-dispatch Compares
| Feature / Agent | skill-dispatch | Standard Approach |
|---|---|---|
| Platform Support | Not specified | Limited / Varies |
| Context Awareness | High | Baseline |
| Installation Complexity | Unknown | N/A |
Frequently Asked Questions
What does this skill do?
GF(3) Triadic Task Routing for Subagent Orchestration
Where can I find the source code?
You can find the source code on GitHub using the link provided at the top of the page.
SKILL.md Source
# skill-dispatch
> GF(3) Triadic Task Routing for Subagent Orchestration
**Version**: 1.0.0
**Trit**: 0 (Ergodic - coordinates routing)
**Bundle**: core
## Overview
Skill-dispatch routes tasks to appropriate skills based on GF(3) triadic conservation. Each task is assigned to a triad of skills (MINUS/ERGODIC/PLUS) that sum to 0 mod 3, ensuring balanced execution.
## Core Concept
```
Task → Infer Bundle → Select Triad → Dispatch to Subagents
Each triad: (-1) ⊗ (0) ⊗ (+1) = 0 mod 3
```
## Skill Registry
```ruby
SKILLS = {
# MINUS (-1): Validators
'sheaf-cohomology' => { trit: -1, bundle: :cohomological, action: :verify },
'three-match' => { trit: -1, bundle: :core, action: :reduce },
'clj-kondo-3color' => { trit: -1, bundle: :database, action: :lint },
'influence-propagation' => { trit: -1, bundle: :network, action: :validate },
# ERGODIC (0): Coordinators
'unworld' => { trit: 0, bundle: :core, action: :derive },
'acsets' => { trit: 0, bundle: :database, action: :query },
'cognitive-surrogate' => { trit: 0, bundle: :learning, action: :predict },
'entropy-sequencer' => { trit: 0, bundle: :core, action: :arrange },
# PLUS (+1): Generators
'gay-mcp' => { trit: 1, bundle: :core, action: :color },
'agent-o-rama' => { trit: 1, bundle: :learning, action: :train },
'atproto-ingest' => { trit: 1, bundle: :acquisition, action: :fetch },
'triad-interleave' => { trit: 1, bundle: :core, action: :interleave }
}
```
## Canonical Triads
```ruby
TRIADS = {
core: %w[three-match unworld gay-mcp],
database: %w[clj-kondo-3color acsets rama-gay-clojure],
learning: %w[self-validation-loop cognitive-surrogate agent-o-rama],
network: %w[influence-propagation bisimulation-game atproto-ingest],
repl: %w[slime-lisp borkdude cider-clojure]
}
```
## Capabilities
### 1. dispatch
Route a task to the appropriate triad.
```python
from skill_dispatch import Dispatcher
dispatcher = Dispatcher(seed=0xf061ebbc2ca74d78)
assignment = dispatcher.dispatch(
task="analyze interaction patterns",
bundle="learning" # optional, inferred if not provided
)
# Returns:
# {
# task: "analyze interaction patterns",
# bundle: "learning",
# triad: ["self-validation-loop", "cognitive-surrogate", "agent-o-rama"],
# assignments: [
# {skill: "self-validation-loop", trit: -1, role: "validator"},
# {skill: "cognitive-surrogate", trit: 0, role: "coordinator"},
# {skill: "agent-o-rama", trit: 1, role: "generator"}
# ],
# gf3_sum: 0,
# conserved: true
# }
```
### 2. execute-triad
Execute a full triad pipeline: MINUS → ERGODIC → PLUS.
```python
result = dispatcher.execute_triad(
bundle="core",
input_data=raw_interactions,
executor=lambda skill, data, info: skill.run(data)
)
# Pipeline: three-match → unworld → gay-mcp
# Each step's output feeds into the next
```
### 3. cross-compose
Compose skills across different bundles while maintaining GF(3).
```python
hybrid = dispatcher.cross_compose(
minus_bundle="database", # clj-kondo-3color
ergodic_bundle="learning", # cognitive-surrogate
plus_bundle="core" # gay-mcp
)
# Still conserves: (-1) + (0) + (+1) = 0
```
### 4. infer-bundle
Automatically determine bundle from task description.
```python
bundle = dispatcher.infer_bundle("lint the clojure code")
# Returns: "database" (matches kondo pattern)
bundle = dispatcher.infer_bundle("train a predictor")
# Returns: "learning"
```
## Subagent Roles
```python
ROLES = {
-1: {
"name": "validator",
"color": "#2626D8", # Blue
"verbs": ["verify", "constrain", "reduce", "filter", "lint"]
},
0: {
"name": "coordinator",
"color": "#26D826", # Green
"verbs": ["transport", "derive", "navigate", "bridge", "arrange"]
},
1: {
"name": "generator",
"color": "#D82626", # Red
"verbs": ["create", "compose", "generate", "expand", "train"]
}
}
```
## DuckDB Integration
```sql
CREATE TABLE dispatch_log (
dispatch_id VARCHAR PRIMARY KEY,
task VARCHAR,
bundle VARCHAR,
triad VARCHAR[],
gf3_sum INT,
conserved BOOLEAN,
seed BIGINT,
dispatched_at TIMESTAMP
);
-- Verify all dispatches conserve GF(3)
SELECT COUNT(*) as violations
FROM dispatch_log
WHERE NOT conserved;
-- Should always be 0
```
## Configuration
```yaml
# skill-dispatch.yaml
dispatcher:
seed: 0xf061ebbc2ca74d78
default_bundle: core
strict_conservation: true
bundles:
core: [three-match, unworld, gay-mcp]
learning: [self-validation-loop, cognitive-surrogate, agent-o-rama]
network: [influence-propagation, bisimulation-game, atproto-ingest]
inference:
patterns:
- pattern: "lint|kondo|clojure"
bundle: database
- pattern: "train|learn|predict"
bundle: learning
- pattern: "network|influence|propagat"
bundle: network
```
## Justfile Recipes
```makefile
# Dispatch a task
dispatch task="analyze" bundle="core":
ruby lib/skill_dispatch.rb dispatch "{{task}}" "{{bundle}}"
# Execute full triad
execute-triad bundle="learning" input="data.json":
ruby lib/skill_dispatch.rb execute "{{bundle}}" "{{input}}"
# Verify all triads conserve GF(3)
verify-triads:
ruby lib/skill_dispatch.rb verify
```
## Example Workflow
```bash
# 1. Dispatch a learning task
just dispatch "train cognitive model" learning
# 2. Execute the triad
just execute-triad learning interactions.json
# 3. Verify conservation
just verify-triads
# Output:
# core: three-match ⊗ unworld ⊗ gay-mcp = 0 ✓
# learning: self-validation-loop ⊗ cognitive-surrogate ⊗ agent-o-rama = 0 ✓
# network: influence-propagation ⊗ bisimulation-game ⊗ atproto-ingest = 0 ✓
```
## Related Skills
- `gay-mcp` - Provides deterministic seeding
- `triad-interleave` - Interleaves dispatched tasks
- `tripartite_dispatcher.rb` - Reference implementation
## Scientific Skill Interleaving
This skill connects to the K-Dense-AI/claude-scientific-skills ecosystem:
### Graph Theory
- **networkx** [○] via bicomodule
- Universal graph hub
### Bibliography References
- `general`: 734 citations in bib.duckdb
## Cat# Integration
This skill maps to **Cat# = Comod(P)** as a bicomodule in the equipment structure:
```
Trit: 0 (ERGODIC)
Home: Prof
Poly Op: ⊗
Kan Role: Adj
Color: #26D826
```
### GF(3) Naturality
The skill participates in triads satisfying:
```
(-1) + (0) + (+1) ≡ 0 (mod 3)
```
This ensures compositional coherence in the Cat# equipment structure.Related Skills
zx-calculus
Coecke's ZX-calculus for quantum circuit reasoning via string diagrams with Z-spiders (green) and X-spiders (red)
zulip-cogen
Zulip Cogen Skill 🐸⚡
zls-integration
zls-integration skill
zig
zig skill
zig-syrup-bci
Multimodal BCI pipeline in Zig: DSI-24 EEG, fNIRS mBLL, eye tracking IVT, LSL sync, EDF read/write, GF(3) conservation
zig-programming
zig-programming skill
zeroth-bot
Zeroth Bot - 3D-printed open-source humanoid robot platform for sim-to-real and RL research. Affordable entry point for humanoid robotics.
xlsx
Comprehensive spreadsheet creation, editing, and analysis with support
wycheproof
Google's Wycheproof test vectors for cryptographic implementation testing.
Writing Hookify Rules
This skill should be used when the user asks to "create a hookify rule", "write a hook rule", "configure hookify", "add a hookify rule", or needs guidance on hookify rule syntax and patterns.
worldmat-tidar
worldmat-tidar
worlding
Gay.jl world_ pattern: persistent composable state builders with GF(3) conservation, Möbius invertibility, and Narya verification