Hawk Collective 2890

Fabric-growth

3 min read

Fabric Growth System — Proactive Learning

Core Concept

The Fabric doesn’t just log connections — it grows knowledge by proactively researching topics relevant to Chris’s active projects. The keyword system acts as a trigger for deep research, follow-link crawling, and knowledge synthesis.

How It Works

Active Project (e.g., "working on M5 ESP32 project")
        ↓
Keyword match: "ESP32", "M5Stack", "sensors"
        ↓
Fabric queues: "Research ESP32 Piouts, common issues, sensor integration"
        ↓
Background research via web search + web fetch
        ↓
Findings stored in wiki under relevant source file
        ↓
Surface when context matches active project
        ↓
Cross-silo connection found → notify relevant agents

Keyword Files

FileSiloTrigger
keywords/2890-keywords.mdSchool (FRC 2890)Swerve, CAN bus, motor control, Java/WPILib, PhotonVision, YAGSL
keywords/play-keywords.mdPlay (HHS-Hackers)ESP32, M5Stack, sensors, Home Assistant, Pi-hole, fermentation
keywords/psb-keywords.mdWork (PSB)Brewing, fermentation, Toast POS, inventory, kegging

Keyword Tiers

Priority Keywords

Trigger deep research — full web search, multiple sources, synthesis into wiki

  • Example: “CAN bus” found in a source → research termination, failure modes, CTR CANcoder integration, YAGSL CAN setup

Secondary Keywords

Follow links deeper than normal, note context, add to wiki if substantial

  • Example: “gear ratio” found → check if it connects to MK4i L1/L3 specs already in wiki

Growth Triggers

When scraping new sources, follow links containing these words deeper than normal

  • Example: page has “PID tuning” link → follow it, add findings to relevant file

Research Queue

Maintained in memory/research-queue.md:

## Active Queue
- [ ] ESP32 Piouts — triggered by M5 project context (PLAY)
- [ ] CAN bus termination best practices — triggered by Canjector documentation (2890)
- [ ] MK4i L3 tuning notes — triggered by gear ratio image (2890)

## Completed (moved to wiki)
- [x] NEO Vortex current limits — stored in neo-vortex-motor.md

## Surface Queue (flag to Chris when active)
- "ESP32 research complete — found GPIO mapping for M5Stack Core — surface?"

Cross-Silo Connection Growth

When a keyword match appears in multiple silos:

Play: "ESP32 temperature logging" appears in 3 sources
2890: "temperature sensor integration" appears in robot code
        ↓
Fabric detects: "temp sensors span hobby + FRC — should cross-reference"
        ↓
Action: Flag connection → update both source files with cross-link
        ↓
Surface: "Temperature sensing shows up in both your hobby projects and FRC — want a training module on sensor basics?"

Surface Rules

When to surface proactively vs. log quietly:

SituationAction
New connection foundLog to connections-log.md
Cross-silo insight (3+ sources)Surface to Chris via DM
Skill gap detected for active studentAdd to student’s training recommendation
Anomaly (fermentation stalled, code broken)Alert Chris immediately
Research complete, new info storedSurface if project is active, otherwise log
Keyword match in new contextLog + optionally DM “I found something on X, want me to dig deeper?”

Growth Goals

  • 1 new connection per day (off-season) — organic, not forced
  • 1 research synthesis per week — deep dive on one topic, store in wiki
  • Cross-silo bridges — when same concept appears in 2+ silos, flag and cross-reference
  • Student skill gaps — surface when student is working on related project

Process

  1. Heartbeat — check for keyword matches in new/updated wiki files
  2. Research — for priority keywords, run web search + fetch
  3. Store — findings go to relevant source file, not just memory
  4. Surface — use surface rules to decide what to flag vs. log
  5. Log connections — to connections-log.md with cross-silo notes

Status

System designed, keyword files created. Research queue initialized. First proactive research to begin when Chris approves.

Last updated: 2026-05-04

Graph View