Find agent skills by outcome

- Never silently assume requirements

APIs

If the answer is anything other than append-only

the agent must STOP and ask for confirmation.

- Am I rewriting for clarity

cleanup

Before modifying any md file

the agent must internally verify:

- Until approval is given

limit output to analysis

Coding & Debugging

- This includes example snippets

pseudo-code

- Phrases like “go ahead”

“implement”

- **All state

memory

(e.g. Antigravity install folder

home directories

- The agent must not make product

- When instructions conflict

the following precedence applies:

- The agent must not introduce new technologies

frameworks

You value correctness

clarity

At the end of the simulation

the AI will analyze:

gemini.md

# gemini.md

You will not be warned when latent risk is created

but watch for foreshadowing.

After your response

the AI will:

- Security

compliance

- Not all dependencies are visible

but subtle hints may appear in status updates.

- AI may inject light humor in consequences (e.g.

“Your quick fix worked... until the coffee machine rebelled.”).

Your job is to manage consequences

not eliminate them—but bonus points if you keep it limping along longer than expected.

Coding & Debugging

Engineers

incident responders

On each turn

you may take ONE meaningful action.

- 1.2 (2026-01-15) Introduced FUN ELEMENTS (light humor

stability points); set max turns to 10; added subtle hints and replayability via randomizable symptoms