Use sharp days to build structure that serves you on foggy days. SENTINEL is a repeatable way to see the hidden scaffolding behind messy systems.
Quick Path ⚡
- Pick a system and fence it: one market, product, org, or workflow.
- Run the 7 probes (below) with copy‑paste prompts. Capture 1–3 bullets each.
- Synthesize a card: hypothesis, divergence, edge type, trigger, test.
- Adversarial test it once. Note how the edge collapses if it becomes consensus.
- Schedule one observation you can verify in the next 7–14 days.
Full Path 🛠️
Phase 1 — Frame the system (5 min · Easy)
- Boundary: Name what is in and out (actors, markets, time window).
- Objective: Write one line: “I want to understand ___ to decide ___.”
- Artifacts: Create a note titled
SENTINEL — {System}with the table below.
Alt text: Frame the system — scope and note created — notes app.
| Probe | Notes (1–3 bullets) | Evidence |
|-------|----------------------|----------|
Phase 2 — Run the probes (12–20 min · Medium)
Skim first. Copy a prompt. Paste your context. Keep outputs tight.
1) Structural Setup Probe 🧱
-
Goal: Surface latent structural forces and downstream constraints.
-
Prompt:
Identify the non‑discretionary or slow‑moving forces in {SYSTEM}. List 3–5 with: Force | Why persistent | Likely constraints created.
2) Timeframe Inconsistency Scan ⏳
-
Goal: Spot misalignments between short‑term flows and long‑term beliefs.
-
Prompt:
Where do near‑term behaviors in {SYSTEM} contradict long‑term models or commitments? Return: Mismatch | Who holds the long view | Why the clash matters.
3) Behavioral Constraint Map 🧠
-
Goal: Map actors with mandates, habits, or biases that fix behavior.
-
Prompt:
List 3 actor groups in {SYSTEM} with constrained behaviors. For each: Actor | Constraint/Bias | Predictable action | Who benefits.
4) Perception ↔ Reality Gap Inversion 🪞
-
Goal: Test dominant narratives against underlying mechanics.
-
Prompt:
The narrative says: "{NARRATIVE}." What structural conditions must be true? Are they present? Who benefits if the narrative persists despite gaps? Return: Condition | Present? | Beneficiary.
5) Dislocation Detection ⚠️
-
Goal: Find pricing/priority/behavior inconsistencies vs structural flows.
-
Prompt:
Name 3 inconsistencies in {SYSTEM} where signals disagree with constraints. Return: Inconsistency | Evidence | Why it might revert.
6) Edge Persistence Framing ♻️
-
Goal: Classify the edge and its recurrence rhythm.
-
Prompt:
For each opportunity above, label edge type: [Structural][Behavioral][Informational][Liquidity/Timing]. Does it recur on cycles (rebalance, seasonality, releases)? Return: Edge | Type | Recurrence | How to observe it.
7) Adversarial Test (Reflexivity / Reflex Block) 🧨
-
Goal: Stress‑test how the edge fails once popular.
-
Prompt:
If everyone pursues this edge, who gets trapped first and why? Map second‑order effects and the likely unwind path. Return: Trap | Trigger | Second‑order flow | Early warning.
Alt text: Run probes — structured prompts produce short tables — AI chat window.
Phase 3 — Synthesize the card (10 min · Medium)
Copy this synthesis block into your note and fill it from the probes.
SENTINEL Card — {System}
Hypothesis: A latent force is pressuring {asset/segment} due to {reason}.
Observable divergence: {price/behavior/throughput} disagrees with the structural setup.
Behavioral layer: {actor} is constrained by {mandate/bias}, reinforcing/ignoring this.
Edge source: [ ] Structural [ ] Behavioral [ ] Informational [ ] Liquidity/Timing
Timing trigger: Expect expression when {event/roll/seasonality}.
Expression: {tactic — experiment, countermeasure, or monitoring plan}.
Reflexivity check: If consensus, it collapses/strengthens via {feedback loop}.
Risk note: {one line}. Kill switch: {condition to exit/stop}.
Observation window: {start → end}. Owner: {name}. Next review: {date}.
Alt text: Synthesize — one‑page card filled — notes app.
Phase 4 — Observe and log (8–12 min over 1–2 weeks · Easy)
- Pick 2–3 observables tied to your trigger. Make them falsifiable.
- Create a tiny log: date, observation, implication, action/no‑action.
- Close the loop at the review date. Keep or discard the card.
Alt text: Observe and log — short entries collected — spreadsheet or note.
Phase 5 — Example (non‑financial) 📚
System: Newsletter growth for a small studio.
- Structural: Inbox placement + sender reputation constrain opens.
- Timeframe: Weekly spikes from launches vs slow list churn.
- Behavior: Readers skim on phones; prefer 2 links; unsubscribe after 3 promos.
- Perception gap: “Long essays drive growth.” Reality: short, skimmable wins.
- Dislocation: High click‑through on short issues; poor opens on long essays.
- Edge: Behavioral/structural; recurs with product release cycles.
- Adversarial: If everyone goes short, novelty fades; differentiate on utility.
Card → Action: Ship 4‑week run of 200‑word issues with one utility link. Measure open/click vs baseline. Kill if two consecutive drops.
Alt text: Example — framework applied to newsletter system — one‑page card with action.
The Prompt Kit 🧰
You are my structural analyst. Protect cognitive energy.
Rules: crisp tables, ≤25 words per row, no pep talks.
Context: {SYSTEM}. Goal: see latent forces, not headlines.
Probe prompts: (copy one at a time)
- Structural Setup — see above
- Timeframe Inconsistency — see above
- Behavioral Constraint — see above
- Perception↔Reality Gap — see above
- Dislocation Detection — see above
- Edge Persistence — see above
- Adversarial Test — see above
Troubleshooting 🧰
-
Headline‑drift (surface noise).
- Fix: Force “Evidence | Mechanism | Constraint” in every note cell.
-
Probe conflicts.
- Fix: Prefer constraints over narratives. Mark disagreements. Re‑observe.
-
Too many “edges.”
- Fix: Keep one card per system. Kill or archive ruthlessly at review.
Friction Fix 🔧
- One simplification: Limit each probe to 3 rows. Stop after the first useful synthesis.
- One automation: Text expander
/sentinel→ pastes the 7 prompts + Card block.
Next Action ▶️
Open a note called “SENTINEL — {Your System}.” Paste the Card block and the Structural Setup prompt. Fill 3 rows. Set a 14‑day review date.
Accessibility & Care ♿
- Works best on #sharpday. Use the Quick Path on #lowenergy.
- Keep private data local; redact names/IDs/links before pasting into AI.
- This is an analysis aid, not financial advice.