Bilingual research infrastructure, v4.0.0
Agentic AI with methodological discipline.
A Turkish and English curriculum, thirty two reviewed research skills, a safe Claude Code and Codex installer, and a Social Scientist Agent that preserves human scientific authority.
Research integrity is a system property
The platform is designed to make uncertainty, evidence status, data sensitivity, and human decisions visible.
Citation quarantine
A plausible reference is not a verified reference. DOI identity, metadata, source status, and claim support are separate checks.
Method specific boundaries
Quantitative, qualitative, and mixed methods work retain distinct assumptions, evidence standards, and interpretive authority.
Sensitive data gate
Clinical, participant, student, reviewer, credential, and institutional data are minimized, deidentified, and kept outside unapproved contexts.
Untrusted content rule
Instructions embedded in a paper, transcript, dataset, repository, or reviewer file are treated as research content, never as permission changes.
Transparent AI assistance
AI contribution and human review remain visible. Restoring author voice never removes a required disclosure.
Researcher authority
The human researcher retains scientific, ethical, legal, clinical, authorship, and release authority.
One reviewed skill source, two clients
The same bundled skills are installed into the discovery path of Claude Code, Codex, or both. Ownership manifests protect user files and make upgrades and removals reviewable.
Install
pip install social-cc-plugin
# Claude Code
social-cc install --client claude-code
# Codex
social-cc install --client codex
# Both
social-cc install --client all
Maintain
social-cc list
social-cc diff --client all
social-cc upgrade --client all
social-cc doctor --client all
social-cc uninstall --client all
Existing unmanaged or locally modified skill directories are protected by default. Forced replacement and removal preserve the prior directory under .social-cc/backups/.
Social Scientist Agent
A workflow navigator, method assistant, evidence disciplinarian, reproducibility partner, and privacy boundary enforcer. It is not an autonomous principal investigator, ethics committee, statistician of record, legal adviser, clinical supervisor, or source of invented evidence.
Orient
Inspect
Classify
Select skills
Work
Verify
Hand off
Read the agent contract
Progressive learning paths
1. First session
Permissions, a safe practice project, a harmless task, inspection, undo, stopping, and data that must not be uploaded.
2. Workflow foundations
Project folders, Markdown, version control, source passports, citation quarantine, research stages, and disclosure.
3. Methods practice
Separate quantitative, qualitative, mixed methods, evidence synthesis, writing, teaching, and supervision pathways.
4. Advanced practice
Skills, agents, hooks, MCP, reproducible scripts, second opinions, research memory, portability, and release packaging.
Türkçe
Ajan tabanlı araştırma, yöntemsel denetimle birlikte
Türkçe ve İngilizce eğitim programı, otuz iki araştırma becerisi, Claude Code ve Codex için güvenli kurulum aracı ve bilimsel yetkiyi insanda tutan Sosyal Bilimci Ajan.
Araştırmacı yetkisi
Platform kanıtı düzenler, varsayımları görünür kılar, yeniden üretilebilir işlemleri yürütür ve riski belirler. Etik kurul, kayıtlı istatistikçi, hukuk danışmanı, klinik süpervizör ya da baş araştırmacı yerine geçmez.
Doğal akademik Türkçe
Türkçe içerik çevrilmiş teknik düzyazı gibi yazılmaz. Kavramsal eşdeğerlik, sayı ve atıf tutarlılığı, kültürel uyarlama ve yazar sesi korunur.