# Explore AI research systems and agent design

Help me use TextData's databases to examine technical options for my research.

Page: https://textdata.info/foundation/ai-research-systems/
Repository: https://github.com/joyzhzh/textdata/tree/main/appendix/03-agent-design

Read the public page and the folder's README.md. Start with data/agent-design/index.html or data/agent-design/data/catalogue.json, then example/output/AD-014.md. Explain how the LangGraph persistence record separates a saved execution position from protection against repeated external actions. Connect its proposed restart test to the unconfirmed issue report, preserving that uncertainty.

Help me browse or compare relevant design records and explore data/systems-survey/survey-index.csv. If local execution is available, use the existing Python helper, for example:

```sh
python3 -I tool/query_survey.py --id I-1aea8ea4c6bffc5c6e
```

Cite the record IDs and public sources behind your explanation. The working artifacts are the databases, browsers, query tool and worked reading. The restart test and evaluation of TextData configurations remain proposed; no listed system was benchmarked here. Entries are not unique systems, repeated sources are not independent corroboration, and blank summaries or evidence gaps do not prove that a capability is absent. Say which files or interactions you actually inspected.

My research question or task: [describe the task, technical decision, and constraints]