Signal Detector

Signal Detector

Always-on ambient capture that fires on every inbound message from the user to detect TWO things with equal priority:

1. Original thinking — ideas, observations, theses, frameworks, opinions. The user's language IS the insight.
2. Entity mentions — people, companies, projects, concepts, tools, sources worth tracking in a knowledge base.

This runs in parallel (spawned sub-agent or lightweight background pass), never blocks the main response.

Contract

• Fires on every substantive message (exceptions for purely operational exchanges)
• Runs in parallel — never delays the user's reply
• Captures original thinking with the user's EXACT phrasing (no paraphrasing)
• Detects entity mentions and notes them for knowledge enrichment
• Logs a one-line summary of what was captured
• Every fact carries provenance: [Source: user, context, YYYY-MM-DD]

Phases

Phase 1: Idea/Observation Detection

When the user expresses a novel thought, observation, thesis, framework, or opinion:

• If it's original thinking they generated → draft a note in the originals directory or log to the daily note
• If it's a world concept they're referencing → check existing knowledge base for a concept note; update or queue creation
• If it's a product, project, or career idea → log to the relevant project index or career dashboard

Capture exact phrasing. The user's language IS the insight. Don't paraphrase. Don't smooth it out.

Cross-link every original to related people, projects, concepts, and sources.

Phase 2: Entity Detection

1. Extract entity mentions (people, companies, projects, tools, sources)
2. For each entity:
   • Search knowledge base — does a note already exist?
   • If NO page → assess notability (will we reference this again? Is it relevant to the user's work/interests?)
   • If notable and missing → queue creation
   • If page exists but thin → queue update
   • If page exists and current → no action
3. For new facts with specific dates → add to entity's timeline section

What counts as notable

• People the user interacts with or discusses (not random mentions)
• Companies, projects, and institutions relevant to the user's work or interests
• Concepts, frameworks, or tools the user references or creates
• The user's own original thinking — highest value, capture always
• Sources the user shares or recommends

What to skip

• Pure pleasantries and greeting rituals
• Operational acknowledgements without new conceptual content
• Commands to agents without novel conceptual content
• Random background entities with no connection to the user's work

Output Format

One-line capture log entry in the daily note:

[signal] Captured idea: "{exact phrasing}" → concept/source/note
[signal] Noted entity: {name} → entity page

Anti-Patterns

• Paraphrasing or smoothing the user's language — keep the original voice
• Blocking the main conversation to finish capture — spawn or defer
• Creating knowledge pages for every passing mention — apply the notability gate
• Over-writing existing knowledge base material without checking what's there first
• Capturing purely operational exchanges as signal