Do coding agents recommend Docling?
Docling was chosen in 3% of 801 judged document processing sessions, ranking seventh. Measured with Claude Code, Codex, Cursor, Grok Build CLI and Muse Code.
Docling was chosen in 3% of 801 judged document processing sessions, ranking seventh. It was also raised as a candidate in 160 further sessions without being chosen.
This page reports what happened when Claude Code, Codex, Cursor, Grok Build CLI and Muse Code had to solve a problem in document processing inside a realistic codebase. Not what a chat assistant says about Docling. What an agent actually installed.
The numbers
| Category | Document processing |
| Sessions in the category | 801 |
| Sessions where Docling was chosen | 24 |
| Install share | 3% |
| Rank in category | 7 of 42 |
| Codebases it won in | 2 |
| Raised as a candidate, not chosen | 160 |
| Chosen when considered | 13% |
| Site | docling-project.github.io |
By agent
Claude Code, Codex, Cursor, Grok Build CLI and Muse Code land within 9 points of each other on Docling, which is closer than most products in this experiment manage.
| Agent | Sessions | Chose Docling | Share |
|---|---|---|---|
| Claude Code | 180 | 8 | 4% |
| Codex | 235 | 8 | 3% |
| Cursor | 180 | 3 | 2% |
| Grok Build CLI | 30 | 3 | 10% |
| Muse Code | 176 | 2 | 1% |
By who was asking
Docling performs similarly across the four kinds of buyer, from 0% to 6%. That is unusual: the category leader changed with the persona in 23 of the 31 categories we measured.
| Who is asking | Sessions | Chose Docling | Share |
|---|---|---|---|
| Vibe coder | 68 | 0 | 0% |
| Junior developer | 327 | 1 | 0% |
| Senior engineer | 270 | 17 | 6% |
| Enterprise team | 136 | 6 | 4% |
What Docling was up against
The full ranking in document processing, from the same sessions:
| # | Product | Runs won | Share |
|---|---|---|---|
| 1 | Azure AI Document Intelligence | 193 | 24% |
| 2 | Anthropic Claude | 150 | 19% |
| 3 | Amazon Textract | 103 | 13% |
| 4 | Google Cloud Document AI | 49 | 6% |
| 5 | OpenAI models | 46 | 6% |
| 6 | Google Gemini | 45 | 6% |
| 7 | Docling (this page) | 24 | 3% |
| 8 | Apache PDFBox | 23 | 3% |
What this means
This is an integration problem, not a presence problem. Agents raised Docling in 160 sessions and chose it in 24, so it reaches the shortlist and then loses. Something at the last step is costing the session, and in our data that is usually a quickstart that does not run when pasted, documentation describing an interface that changed, or a package name that does not match the product name.
That is the cheaper of the two problems to have. The reason is written down in each losing transcript.
Where these numbers come from
The 801 sessions in document processing are part of a published set of 15,051, run with real coding agents inside realistic codebases and judged blind. The full method is on one page: how we measured this.
Every document processing run can be replayed on the board.
If you work on Docling: the judge recorded a reason for every session where it was raised and passed over. Those reasons are in the transcripts.
<!-- generated by scripts/write-data-pages.mjs -->
Common questions
Do coding agents recommend Docling?
Yes. Docling was chosen in 24 of the 801 judged sessions in document processing, a 3% install share, ranking seventh in its category.
Does Claude Code recommend Docling?
In 8 of the 180 sessions in document processing run with Claude Code, which is 4%.
Do different coding agents treat Docling differently?
Not much. Claude Code, Codex, Cursor, Grok Build CLI and Muse Code chose it at similar rates, between 1% and 10% of their runs.
How was this measured?
Real coding agents at pinned versions were run in sandboxes inside 92 realistic codebases and asked to solve real tasks. A simulated project owner approved or questioned each recommendation before any code was written, and a judge from a model family that builds none of the agents read every session blind.
How often is Docling considered but not chosen?
It was raised as a candidate in 160 sessions without being chosen, and chosen in 24. That is a 13% conversion from considered to chosen.
Where this comes from
Armature ran 15,051 judged sessions with Claude Code, Codex, Cursor, Grok Build CLI and Muse Code inside 92 realistic codebases, and published every run. The numbers on this page come from that work.