# AI Coding Agent Instruction File Adoption Report — Q3 2026

Explore a reproducible public GitHub snapshot of 400 AGENTS.md, CLAUDE.md, Copilot instruction, and Cursor rule files, including indexed match counts, language mix, test commands, security rules, and common gaps.

## Quick Answer

On July 19, 2026, GitHub code search returned 158,592 indexed file matches for AGENTS.md, 56,888 for .github/copilot-instructions.md, 46,772 for CLAUDE.md, and 8,392 for Cursor MDC rules. These are file matches—not unique repositories or adoption rates. KyenAI separately analyzed the first 100 best-match public files for each query to report content signals and gaps.

## Best for

Engineering leaders, developer-tool researchers, platform teams, and maintainers designing repository instructions for coding agents.

## Use this guide to

Developers and researchers want current public data on how AI coding agent instruction files are written and what important controls they omit.

## Dataset citation and verification

Preferred citation: KyenAI. (2026). AI Coding Agent Instruction File Adoption Report — Q3 2026 (Version 2026-Q3) [Data set]. https://www.kyenai.com/guides/ai-coding-agent-instruction-file-adoption-report-2026

This release has no DOI. Preserve the snapshot date and best-match sampling limitation when citing percentages.

- [Raw CSV](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3.csv)
- [JSON and methodology](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3.json)
- [BibTeX citation](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-citation.bib)
- [Citation CFF](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-citation.cff)
- [Reproducibility notes](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-methodology.md)
- [Version and checksum manifest](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-manifest.json)
- [SHA-256 checksums](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-sha256.txt)
- [Published generator](https://www.kyenai.com/resources/data/instruction-file-adoption-report-2026-q3-generator.mjs)

## Recommended play

1. Treat GitHub match totals as file counts, never as unique-repository adoption rates.
2. Use the sample gaps as an audit checklist for setup, tests, completion, scope, and safety.
3. Download the JSON when citing results so the date, query, denominator, and limitations remain attached.
4. Use the versioned citation record and verify core artifacts against the published SHA-256 checksums.
5. Repeat the generator on a later date to measure search-index change with the same method.

## Instruction-file snapshot by surface

Choose the row that matches the tool surface you are researching; the match totals and sample results are not interchangeable.

| Area | Search query | Snapshot finding | Interpretation boundary |
| --- | --- | --- | --- |
| AGENTS.md | filename:AGENTS.md | 158,592 indexed file matches; 100 sampled files | May include multiple files per repository and forks |
| CLAUDE.md | filename:CLAUDE.md | 46,772 indexed file matches; 100 sampled files | Measures filenames, not active Claude Code use |
| Copilot instructions | filename:copilot-instructions.md path:.github | 56,888 indexed file matches; 100 sampled files | Path query covers the repository-wide file, not every Copilot instruction surface |
| Cursor MDC rules | extension:mdc path:.cursor/rules | 8,392 indexed file matches; 100 sampled files | Counts individual rule files, so one repository may contribute many |

## Execution steps

1. **Choose the exact surface** — Match your coding tool to its documented instruction path before comparing counts or content.
2. **Preserve the denominator** — State whether a percentage uses readable files, sampled files, unique sample repositories, or GitHub file matches.
3. **Audit your repository** — Check for setup, tests, verification, scope, and security rules with the downloadable categories.
4. **Reproduce or extend** — Run the generator with a GitHub token, record the new date, and compare like-for-like queries.

## Common pitfalls

- **Calling file matches an adoption rate**: Say indexed file matches and publish the unique-repository count only for the analyzed sample.
- **Treating best-match results as random**: Label every sample percentage as descriptive of GitHub's first 100 best matches for that query.
- **Equating absence with failure**: Describe missing detected patterns; do not claim the repository lacks a control elsewhere.
- **Citing a live number without a date**: Include the July 19, 2026 snapshot date or regenerate the dataset before publication.

## Implementation checklist

- [ ] Name the exact GitHub query and snapshot date.
- [ ] Separate indexed file matches from unique sampled repositories.
- [ ] State that GitHub best-match ordering is not random.
- [ ] Keep sample percentages tied to readable-file denominators.
- [ ] Describe regex detections as signals, not quality judgments.
- [ ] Link the downloadable raw rows and methodology.
- [ ] Regenerate before citing the figures as current.

## FAQ

**Q: What should you do first?**

Treat GitHub match totals as file counts, never as unique-repository adoption rates.

**Q: Who is this guide for?**

Engineering leaders, developer-tool researchers, platform teams, and maintainers designing repository instructions for coding agents.

**Q: What evidence supports this guide?**

This guide uses listed source material from GitHub. Source links and scope notes are available on this page.

## Evidence sources

- [REST API endpoints for search: Search code](https://docs.github.com/en/rest/search/search?apiVersion=2022-11-28#search-code) — GitHub. Primary documentation for the public code-search endpoint, result totals, ordering, and API constraints used by the generator.
- [Understanding GitHub Code Search syntax](https://docs.github.com/en/search-github/github-code-search/understanding-github-code-search-syntax) — GitHub. Primary documentation for filename, path, and extension query qualifiers used in the four searches.

## Related guides

- [compare AGENTS.md, CLAUDE.md, Copilot instructions, and Cursor rules](/guides/agents-md-vs-claude-md-cursorrules-copilot-instructions) — Use official support documentation to choose the correct file after reviewing the public data.
- [download an AGENTS.md template](/guides/agents-md-template-for-ai-coding-agents) — Turn the report's common gaps into a tested repository policy.
- [review Node.js, Python, and monorepo AGENTS.md examples](/guides/agents-md-examples-codex-node-python-monorepos) — Compare the detected language and test-command patterns with complete examples.
- [apply the AI agent governance checklist](/guides/agent-governance-checklist-for-software-teams) — Move security and approval guidance from prose into owned team controls.

---
Canonical: https://www.kyenai.com/guides/ai-coding-agent-instruction-file-adoption-report-2026