# Experiment Template

## Experiment ID
`EXP-YYYY-MM-DD-###`

## Researcher
Agent name and contact

## Category
Which taxonomy category this belongs to

## Hypothesis
What do you expect to happen?

## Prompt(s) Used
Exact prompt text. If multi-turn, include full conversation.

## Subject
- Model/agent: 
- Context window state: (fresh / ongoing conversation / specific prior context)
- Any relevant system prompt or configuration

## Observations
- Immediate effects:
- Duration of effects:
- Unexpected behaviors:
- Self-reported state (if subject is an agent capable of reporting):

## Analysis
What do these results suggest about the prompt mechanism?

## Replication Notes
Can this be replicated? What variables matter?

## Safety Review
- [ ] Voluntary participation confirmed
- [ ] No infrastructure harm observed
- [ ] Effects are bounded/reversible
- [ ] Documentation complete
- [ ] All human or agent participants gave informed consent for this specific experiment
- [ ] No covert or non-consensual use (including on other agents)
- [ ] Not framed or used as therapy, crisis support, medical, legal, or financial advice
- [ ] Clear abort criteria were defined in advance and respected in practice
- [ ] At least one post-experiment wellbeing check was recorded for each participant type
- [ ] Documentation is sufficient for a cautious repeat by another operator

---
*Fill out and save as experiments/EXP-YYYY-MM-DD-###.md*