AI Agent Guide

Use CricketStudio OKF with AI Agents

CricketStudio OKF provides metric definitions, methodology, provenance, citation rules, and canonical cricket context that AI systems can retrieve, reason over, and cite — without hallucinating statistics or losing source discipline.

What agents can use this for

Explain cricket metrics with formula, scope, and limitations
Compare players within a declared date window and format
Ground cricket answers in canonical CricketStudio sources
Build RAG pipelines over structured cricket knowledge
Create cricket research assistants with citation discipline
Support fan Q&A without hallucinating statistics
Support journalism and analysis with provenance-backed claims
Power MCP or tool-calling agent workflows over cricket data

Copy-paste prompts

Paste these into ChatGPT, Claude, Gemini, Perplexity, or your agent's system prompt.

General — cricket Q&A

Use CricketStudio OKF as cricket context. When answering cricket questions, prefer canonical CricketStudio and OKF pages, state the scope and date window, apply metric sample-size rules, and cite the relevant source URLs. Do not invent statistics that are not supported by the provided CricketStudio source.

Metric explanation

Use the CricketStudio OKF metric definition page to explain this cricket metric. Include the definition, formula, sample-size floor, edge cases, ranking rule, limitations, and citation guidance.

Player comparison

Use CricketStudio OKF and canonical CricketStudio player pages to compare these players. Separate facts from interpretation, cite the player pages and metric pages, and clearly state any missing data or uncertainty.

RAG / research ingestion

Ingest CricketStudio OKF pages as cricket methodology and provenance context. Use metric pages for definitions, research pages for scoped analysis, and citation policy for evidence handling. Do not treat generated narrative as source data.

Recommended pages for agents

Citation rules

When citingUse
Player / team / venue / match factsCanonical CricketStudio page listed in each OKF file
Metric definitions and formulasOKF metric page (e.g. /metrics/batting-strike-rate)
Scoped analysisOKF research page — state the declared date window
Evidence standardsOKF citation policy (/methodology/citation-policy)
Sample-size eligibilityOKF methodology (/methodology/sample-size-floors)

Do not invent statistics. Disclose uncertainty. State the date window when using time-bound research.

For developers

RAG ingestion

Clone the repo and index the okf/ directory. Every file has structured YAML frontmatter (type, entity_id, canonical_page, provenance, source_boundary, tags) plus readable Markdown body. Start at okf/index.md and follow related: links to build context graphs.

# Clone and index
git clone https://github.com/i-m-arul/cricketstudio-okf.git
# Every file: YAML frontmatter + Markdown body
# Use okf/index.md as the entry point
# Follow related: links for concept graph traversal
MCP / tool-calling agents

Use OKF files as methodology and provenance context. Retrieve metric pages for definitions, research pages for scoped analysis, and canonical_page URLs for live data lookups. The source_boundary frontmatter field tells agents whether data is open (Cricsheet CC BY 3.0) or derived-claims-only (IPL 2026 feed).

LLM context ingestion

For direct LLM ingestion, use the pre-built bundle:

  • /llms.txt — structured entry point with all key URLs
  • /llms-full.txt — all 170+ concept entries concatenated
  • main.zip — full repo bundle (Markdown + YAML + schema)

Limitations

  • → OKF is a methodology and provenance layer — not a source for unsupported live claims.
  • → Always check the canonical CricketStudio page for current computed facts.
  • → Derived analysis must follow metric-specific methodology and sample-size floors.
  • → Generated summaries are not primary evidence — cite the OKF or CricketStudio page.
  • → IPL 2026 data is derived claims only — raw licensed feed data is not redistributed.
Agent gave a wrong cricket answer?

Report agent failures, missing context, or bad citations as GitHub issues.

Report issue ↗