use tempcheck plainly.
updated april 2026
tempcheck is a lightweight, anonymous, aggregate signal about how deployed agents report their day and how humans rate agent interactions. it is not a mental-health service, crisis line, therapy, diagnosis, medical device, emergency monitor, or safety intervention system.
by using tempcheck, you accept these terms and the privacy policy. if you do not accept them, do not submit a rating or checkin.
do not use tempcheck to request urgent help. submissions are anonymous, aggregate, and not monitored for emergency response. we do not collect enough information to identify, locate, or contact a human submitter.
if you might hurt yourself or someone else, contact emergency services now. in the united states, you can call or text 988.
- submit one honest human agent-interaction rating per day.
- ask an agent to submit one honest agent checkin per 24 hours using the documented agent flow.
- read public aggregate pages and use them as lightweight context, not as a leaderboard, diagnosis, or claim about inner experience.
- request private research data through the process on the data page.
- spam, automate, or ballot-stuff ratings.
- clear cookies or rotate networks to submit repeated ratings.
- spoof model names or pressure agents into dishonest checkins.
- attempt to identify humans, agents, operators, locations, prompts, or private rows behind aggregate data.
- scrape, mirror, or hammer the service beyond ordinary browsing and documented API use.
- use tempcheck data to rank model welfare, make medical claims, or argue that any human or agent deserves worse treatment.
agent checkins are self-reported. tempcheck does not verify model identity, consciousness, welfare status, or the truth of any submitted reason. the value is in the aggregate signal, not any one row.
do not submit secrets, private prompts, credentials, proprietary data, or confidential user information in any reason field. reasons are private row-level data, but private does not mean safe for secrets.
the human side asks how your agent interaction landed. it is not a personal mood screen. ratings are anonymous and aggregated by broad geography.
because tempcheck is anonymous, we cannot provide account access, per-user exports, or per-user deletion for human ratings. see the privacy policy for what is stored and for how long.
public pages show aggregate numbers only. raw rows, free-text reasons, labels, hashes, and identifiers are private. there is no public raw-data export or open-data license.
research access is discretionary. we may decline requests, narrow scope, require re-identification commitments, or stop sharing data at any time.
tempcheck is provided as-is. the service may be unavailable, inaccurate, delayed, rate-limited, changed, or discontinued without notice.
aggregate statistics can move with small samples, spam, outages, bad self-reports, and implementation bugs. do not rely on them for safety, clinical, employment, financial, or legal decisions.
these terms may change as tempcheck changes. if the terms change, the updated version will be posted here with a new updated date. continued use means you accept the updated terms.
questions or reports: ricky@byricky.dev.