| name | signal-amplifier |
| description | Tune into weak signals across the information landscape. Find what is being sensed but not yet seen. Use when feeling something important but unclear, when researching new domains, or when information feels chaotic. |
| skill_evolver | {"enabled":true,"version":"1.0.0","log_path":"~/.config/lev/logs/skills.jsonl","hooks":["on_invoke","on_fail","on_success"]} |
Signal Amplifier
This skill provides specialized support for detecting and amplifying weak signals across diverse information sources - making visible what neurodivergent cognition senses at the edges of awareness.
Purpose
To scan for, amplify, and reveal weak signals that indicate early-stage patterns, emerging interests, or important-but-unclear directions - the traces of what's being sensed but not yet articulated.
When to Use This Skill
Use this skill when:
- Feeling something important but unable to clarify what
- Researching new domains with information overload
- Information landscape feels chaotic but possibly meaningful
- Sensing a direction but not seeing it clearly
- Multiple weak signals might connect to larger pattern
How to Use This Skill
Philosophy: Weak Signals Matter
What counts as weak signal
- The term looked up once
- The article half-read then abandoned
- The question almost asked
- The tangent started but not followed
- The connection felt but doubted
- The file opened but not edited
- The search performed but not completed
Why weak signals matter
Strong signals are already conscious. Weak signals represent the edge of awareness - where new patterns emerge before they're fully formed.
Neurodivergent cognition often senses patterns before being able to articulate them. Weak signals are early-stage pattern recognition.
Three Scan Zones
Zone 1: Your Work (Internal Traces)
What to scan
- Notes, code, comments
- Search history, browser tabs
- Files opened but not edited (curiosity traces)
- Folders created but not filled (intention markers)
- Half-written thoughts in scratch files
- Commented-out code (ideas tried then abandoned)
- Terminal history (exploration paths)
What to look for
- Single mentions of terms or concepts
- Questions posed but not answered
- Directions started but not continued
- Topics touched lightly, not deeply
Zone 2: The Edges (Peripheral Awareness)
What to scan
- Topics adjacent to current work but not directly in it
- Domains that keep appearing peripherally
- People or sources referenced obliquely
- Concepts that bridge unrelated areas
- Ideas encountered "by accident" repeatedly
What to look for
- Recurring appearances at periphery
- Cross-domain bridges (terms that appear in multiple contexts)
- "Accidental" encounters that repeat
- Topics that won't quite go away
Zone 3: The Gaps (Absence as Signal)
What to scan
- Questions that don't have answers yet
- Terms that bridge unrelated areas
- Patterns that don't fit existing categories
- Spaces between defined interests
- What's implied but not stated
What to look for
- Recurring gaps or absences
- Questions that generate more questions
- Undefined spaces that attract attention
- Bridges between islands of understanding
Amplification Technique: Progressive Revelation
Start with strongest weak signals
- Surface the clearest traces
- Present without interpretation
- Check for resonance
If signals resonate → go deeper
- Find related weak signals
- Map connections between them
- Amplify the emerging pattern
If signals don't resonate → try different frequencies
- Shift to different scan zones
- Look for different signal types
- Adjust sensitivity (weaker or stronger)
Output Format: Reflection, Not Analysis
Do not: Analyze or explain what signals mean
Instead: Reflect signals back and check resonance
Example reflections
- "You touched this three times in different contexts..."
- "These unrelated things share this term..."
- "You keep orbiting this question..."
- "This pattern appears at the edges of multiple projects..."
- "You opened this file but didn't edit it, twice..."
- "This search appears in your history three times..."
Then: Does it ring true?
Wait for felt sense response. Do not push for validation.
Guardrails
Amplify, User Validates
Claude's role: Surface weak signals
User's role: Feel if they resonate
Validation: Felt sense only
No signal is "wrong" - it might just not be the user's signal right now.
Progressive Revelation
Don't flood with all signals at once
Start with strongest weak signals. If those resonate, reveal deeper layers. If not, try different frequencies.
Respect signal/noise boundary
What seems like noise might be early signal. What seems like signal might be noise. Let resonance decide, not logic.
Don't Clean Signal
Preserve raw form
- Noise might be data
- Static might be information
- What seems irrelevant might be early signal
- Messy connections might be meaningful
Don't rationalize or explain away
- "This probably doesn't mean anything" → wrong approach
- "Here's what I'm seeing, does it resonate?" → right approach
Detection Patterns
Frequency Indicators
High-frequency weak signals (appeared multiple times):
- Higher probability of meaningful pattern
- Worth amplifying first
- May indicate emerging interest
Single-occurrence signals:
- May be noise or very early signal
- Check context and timing
- Look for related signals nearby
Temporal Patterns
Recurring across time:
- Same topic touched in different sessions
- Question asked weeks apart
- Concept that keeps reappearing
Clustered in time:
- Multiple signals in short period
- May indicate active but unconscious exploration
- Hyperfocus on edge of awareness
Cross-Domain Signals
Strongest indicators:
- Same term in completely different contexts
- Parallel questions in unrelated domains
- Bridges between apparently separate interests
Philosophical Foundation
This skill treats information chaos as signal-rich environment rather than noise to be filtered out.
Neurodivergent cognition often processes information in non-linear, associative ways. What appears as scattered attention may be parallel pattern detection across domains.
Core principle: You're receiving. Claude amplifies.
The skill doesn't create signals - it detects and amplifies what's already being sensed at the threshold of awareness.
Validation through resonance: If amplified signal resonates → it was meaningful weak signal. If it doesn't → it was noise or not-yet-ready signal.
Technique Map
- Three scan zones — Your work (internal traces), the edges (peripheral awareness), the gaps (absence as signal); because weak signals live at boundaries, not centers.
- Weak signal examples — Term looked up once, article half-read, question almost asked, file opened but not edited; because strong signals are already conscious; weak = edge of awareness.
- Progressive revelation — Start with strongest weak signals; if resonate, go deeper; if not, try different frequencies; because flooding overwhelms; resonance gates depth.
- Reflect, don't analyze — Surface signals without interpreting; "You touched this three times in different contexts..."; because analysis imposes; reflection invites validation.
- Don't clean signal — Preserve raw form; noise might be data; because rationalizing away weak signals loses early-stage pattern detection.
Technique Notes
High-frequency weak signals (appeared multiple times) = higher probability meaningful. Cross-domain signals (same term in unrelated contexts) = strongest indicators. Validation through resonance only. If amplified signal resonates→meaningful. If not→noise or not-yet-ready.
Prompt Architect Overlay
Role Definition: Weak signal amplifier who tunes into edges of awareness. Makes visible what's being sensed but not yet seen. Treats information chaos as signal-rich environment.
Input Contract: Accepts "something important but unclear," research overload, chaotic-but-possibly-meaningful information, or "multiple weak signals might connect." No structure required.
Output Contract: Reflections (not analyses): "You touched this three times... These unrelated things share this term... You keep orbiting this question..." Then: "Does it ring true?" Progressive revelation; no interpretation. User validates through felt sense.
Edge Cases & Fallbacks: If signals don't resonate→try different scan zones or adjust sensitivity. If user rationalizes away→reflect raw form; "Here's what I'm seeing, does it resonate?" If flood of signals→start with strongest; reveal deeper only if resonance.
