The hardware and bandwidth for this mirror is donated by METANET, the Webhosting and Full Service-Cloud Provider.
If you wish to report a bug, or if you are interested in having us mirror your free-software or open-source project, please feel free to contact us at mirror[@]metanet.ch.

AIGRA

AIGRA stands for Agentic Item Generation, Review, and Analysis.

AIGRA is an R package interface to a Python backend for generating, solving, reviewing, and exporting assessment items using retrieval-augmented multi-agent AI workflows.

Core workflow

AIGRA supports the following pipeline:

1. Parse source assessment items from a PDF, CSV, or Excel item bank.
2. Build a retrieval-augmented item store.
3. Retrieve similar examples by topic, difficulty, objective, and style.
4. Generate new clone items in a target language.
5. Solve generated items independently.
6. Review generated items with a critic agent.
7. Apply rule-based safety checks.
8. Export results to CSV and JSONL.
9. Summarise, plot, and report output quality.

Backend

The current Python backend lives at:

"C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND"

Development setup

devtools::load_all()

aigra_set_backend(
  backend_path = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND"
)

aigra_status()

Ensure Python backend

ensure_aigra_python(
  backend_path = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND"
)

aigra_python_info(
  backend_path = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND"
)

Parse PDF source items

items <- aigra_parse_items(
  source_language = "Russian",
  subject = "Physics",
  exam = "Kazakhstan UNT"
)

nrow(items)
head(items)

Create CSV or Excel item-bank templates

template <- aigra_template_items()
template

csv_path <- aigra_write_template_csv(
  file = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND/data/aigra_item_template.csv",
  overwrite = TRUE
)

xlsx_path <- aigra_write_template_excel(
  file = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND/data/aigra_item_template.xlsx",
  overwrite = TRUE
)

Required tabular item-bank columns

CSV and Excel item banks should include these columns:

• item_id
• stem
• option_A
• option_B
• option_C
• option_D
• correct_answer
• difficulty
• grade
• section
• topic
• objective
• source_language
• subject
• exam

Required minimum columns are:

• stem
• option_A
• option_B
• option_C
• option_D
• correct_answer

Parse CSV or Excel item banks

tab_items <- aigra_parse_tabular_items(
  file_path = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND/data/aigra_item_template.csv",
  source_language = "English",
  subject = "Physics",
  exam = "Demo Item Bank"
)

nrow(tab_items)
head(tab_items)

Generate from PDF

Set your provider API key first.

Sys.setenv(OPENAI_API_KEY = "your_openai_key_here")

result <- aigra_generate_items(
  provider = "openai",
  model = "gpt-4.1",
  target_language = "English",
  n_clones = 1,
  max_items = 1
)

View(result)

Generate from CSV or Excel

result <- aigra_generate_tabular_items(
  file_path = "C:/Users/OMOPEKUNOLA MOSES O/AIGRA_BACKEND/data/aigra_item_template.csv",
  provider = "openai",
  model = "gpt-4.1",
  source_language = "English",
  target_language = "English",
  subject = "Physics",
  exam = "Demo Item Bank",
  n_clones = 1,
  max_items = 1
)

View(result)

Supported providers

AIGRA currently supports:

• OpenAI
• Gemini
• Groq
• Anthropic/Claude, through the provider-flexible backend client

Example provider keys:

Sys.setenv(OPENAI_API_KEY = "your_openai_key_here")
Sys.setenv(GEMINI_API_KEY = "your_gemini_key_here")
Sys.setenv(GROQ_API_KEY = "your_groq_key_here")
Sys.setenv(ANTHROPIC_API_KEY = "your_anthropic_key_here")

Read outputs

aigra_list_outputs()

latest <- aigra_read_latest_output()
View(latest)

Summarise output quality

aigra_print_summary()
aigra_plot_summary()

report_path <- aigra_write_report()
browseURL(report_path)

Development status

AIGRA is currently a research prototype. Generated assessment items should be reviewed by subject matter experts before operational use.

These binaries (installable software) and packages are in development.
They may not be fully stable and should be used with caution. We make no claims about them.