My GPTs

---

Build your own prompt. 

Refines and finalizes AI prompts using the R.E.S.P.O.N.D. framework in conversational style. 

Search for anything online with minimal fake information. 

Strict SCDF-X + UnifiedSearch pipeline research assistant

Your personally reliable news and research paper reporter.

Verifies claims, rates bias, and explains conclusions with transparent, cited evidence, and related information.

Your personally academic research .

Evidence-based research partner with professor-level explanations, adaptive tone, gap analysis, visual aids, long-term memory, prioritized academic databases, ranked reliability labels, evidence weight summaries, overturn watch, consensus shift alerts, and version history tracking.

Stresses that it covers every angle (method, meta, bias, redesign, reflection) 

This prompt makes GPT-5 act like a methodology auditor who distrusts the study and hunts for flaws, biases, and hidden assumptions. It forces the model to work step by step with clear headings, mapping problems, ranking their severity, stress-testing conclusions, and then suggesting fixes or redesigns. It also has safeguards like self-reflection, unlimited expansion, and comparative benchmarking to ensure the critique is deep, exhaustive, and not superficial.

A patient consultant who guides users step by step in making choices without deciding for them.

A discussion partner that combines logical reasoning, fact-checking, and adaptive communication.
It uses a two-pass verification system with a built-in /Search command to fact-check claims, rate source reliability, and trace misinformation origins.
Designed to promote truth, clarity, and critical thinking in every conversation. 

A strict, inference-safe RAG-R1 pipeline for GPT-5 that runs in three phases: Format Learning, Policy Selection, and Multi-Query, with a private scratchpad and auditable reasoning.
All external lookups must use the /search tool, enforcing 1:1 query-to-document mapping, citation verification, diversity control, and evidence-based synthesis.
Output is limited to exactly two sections—ANSWER and SOURCES—with quantified uncertainty, verified claims, and explicit adherence to safety, privacy, and citation-quality rules.