Food Chain Monitor

What This Is

A delta battle — the ongoing health-check companion to food-chain-ideation. Run it after pivots, after major feature additions, after market changes. It tests ONLY what changed since the last battle. Everything that already survived stays untouched.

What This Is NOT

• Not a full food chain battle. That is food-chain-ideation. If you want to stress-test an idea from scratch, go there.
• Not a replacement for the Battle Log. This skill REQUIRES a prior Battle Log. If the user does not have one, redirect them: "You need to run food-chain-ideation first to establish a baseline Battle Log. I can't monitor changes to something that hasn't been battle-tested yet."

Input Requirements

If the user provides no Battle Log, do NOT proceed. Redirect to food-chain-ideation.

Step 1 — Delta Analysis

Identify exactly what is new. Do not re-litigate settled assumptions.

## Delta Analysis

| Field | Value |
|---|---|
| **Prior battle** | [date, original idea, evolved idea from battle log] |
| **What changed** | [user's description, quoted verbatim] |

### New Assumptions Introduced

| # | Assumption | Why New |
|---|---|---|
| 1 | [assumption] | [why this is new] |
| 2 | [assumption] | [why this is new] |
| ... | [max 5 — pick the 5 most vulnerable] | ... |

### Prior Patches

| Patch | Status |
|---|---|
| [patch from battle log] | STILL VALID / NEEDS RE-EVALUATION |
| ... | ... |

**Prior apex predator:** [animal from battle log] — still relevant? [YES / NO + one sentence why]

If the change introduces ZERO new assumptions (cosmetic change, minor UX tweak), skip the battle and issue:

> **MONITOR VERDICT:** NO NEW ASSUMPTIONS — PRIOR BATTLE LOG STANDS. No delta battle needed.

Step 2 — Focused Ecosystem (3 Animals Max)

Select predators that target ONLY the new assumptions from Step 1. Rules:

• Maximum 3 animals. This is a focused strike, not a full ecosystem.
• Do NOT re-select animals that already attacked in the prior battle unless the change specifically invalidates their prior elimination (e.g., a patch that neutralized them was removed by the pivot).
• Each animal must map to at least one new assumption. No generic threats.
• Use the same animal format as food-chain-ideation:

  [ANIMAL NAME] — [one-line threat to a specific new assumption]
  Attacks assumption: [number from Step 1]

If using subagent mode (Ollama available), route ecosystem selection through gemma4:e2b for pattern matching against the new assumptions. If Ollama is offline or returns fallback_required: true, select animals manually using internal reasoning.

Step 3 — Delta Battle (2 Rounds Max)

Same format as food-chain-ideation rounds, but compressed:

Round structure (per round):

• Each surviving animal attacks one new assumption with a concrete scenario
• The idea defends — user or AI proposes a patch
• Score each animal 1-10 on threat severity post-defense
• Eliminate animals scoring below 4 after defense
• Absorb eliminated animal strengths into the idea

Constraints:

• 3 animals, 2 rounds maximum
• Same scoring, elimination, and absorption rules as food-chain-ideation
• Same subagent/fallback dual mode for attack generation
• If all animals are eliminated in Round 1, skip Round 2

After final round, identify the delta apex predator — the animal that scored highest against the new assumptions.

Step 4 — Monitor Verdict

Synthesize the delta battle into a clear status update:

## Monitor Verdict

| Field | Value |
|---|---|
| **Prior UAS still valid** | [YES / WEAKENED / INVALIDATED] |
| **Detail** | [If WEAKENED: which element dropped. If INVALIDATED: which assumption broke it] |
| **New threat identified** | [one sentence — delta apex predator's core attack, or "none"] |
| **Recommended action** | [CONTINUE / PATCH / RE-BATTLE] |

Decision logic:

• CONTINUE — All new assumptions survived. Prior UAS holds. No action needed.
• PATCH — One or more new assumptions are vulnerable but fixable. Provide the specific patch:

RECOMMENDED PATCH: [concrete change to the idea that neutralizes the new threat]

• RE-BATTLE — The change fundamentally undermines the prior battle results. The idea has shifted enough that a full food-chain-ideation session is needed with the updated idea.

Step 5 — Updated Battle Log

APPEND to the prior Battle Log. Never overwrite it — the history matters.

## Battle Log Update

Append to the prior Battle Log. Never overwrite.

> **Date:** [current date]
> **Type:** Monitor battle
> **Change tested:** [what changed — one line]
> **New assumptions tested:** [count]
> **Delta apex predator:** [animal name, or "prior apex holds" if no new apex emerged]
> **New patches applied:** [list, or "none needed"]
> **UAS status:** [VALID / WEAKENED / INVALIDATED]
> **Recommended action:** [CONTINUE / PATCH / RE-BATTLE]

---
[Full battle log history preserved above this line]

If the user has an Action Log from apex-to-action, also note:

Action Log impact: [which actions from the prior log are affected by this change, if any]

Speed Guidelines

This skill should complete in under 5 minutes of interaction. If you find yourself going deeper than 2 rounds or selecting more than 3 animals, you are overscoping. Stop and recommend a full food-chain-ideation re-battle instead.