docs(agents): survey Posit's AI tooling repositories (closes #106) - #211
Conversation
The lab already adopted posit-dev/skills but had no survey of the rest of the organization, so questions like "should we use btw or mcptools with Claude Code" were answered from memory. This adds a chapter fragment that enumerates the organization from the GitHub API (226 public repos on 2026-09-09), tabulates the 16 that build or evaluate LLM tooling, and gives an R-centric verdict for each, singling out btw, mcptools, and mcp-repl as worth adopting and bluffbench2 as a reference eval. It also notes that ellmer and ragnar live under tidyverse and that Positron Assistant has no standalone public repo, since both are common expectations that the listing cannot satisfy. The section is transcluded right after the Posit skill collection so the two Posit sections sit together, and each table row cites its repository in references.bib. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
There was a problem hiding this comment.
Copilot wasn't able to review any files in this pull request.
💡 Add a code-review agent skill or configure MCP servers for context-aware, tailored reviews. Learn more in the docs.
An adversarial review of the posit-dev survey found prose that
contradicted its own table and two claims stated more firmly than the
evidence allows. The table verdict for mcptools now matches the prose
(it is the layer btw uses, not itself the session inspector); commons is
no longer lumped in with the Shiny chat components, since it is a data
agent framework; the ellmer sentence now distinguishes the packages that
build on it from those that integrate with it; the Positron Assistant
architecture is attributed to the repository descriptions; and the
Rscript -e cost is stated as the mechanism (a fresh process per call)
rather than as an observation. Filler verdicts ("worth trying", "worth
knowing") are replaced with direct ones, the section heading counts
tools rather than packages because mcp-repl is a Rust binary, and
ellmer, ragnar, and vitals get repository links.
Refs #106.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
|
Pushed with the pre-push guard's Posted by Claude Code (AI agent) --- not written by a human. |
There was a problem hiding this comment.
🟡 Changes recommended
The newly added references.bib entries don’t follow the file’s established howpublished = {Software} convention for GitHub repositories.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Review details
- Files reviewed: 4/4 changed files
- Comments generated: 1
- Review effort level: Lite
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
…tion
Copilot flagged the 17 new posit-dev repository entries for using
`howpublished = {GitHub repository}` where the file uses `{Software}`
for GitHub repositories; they now match.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
🟢 Approval recommended
The changes are documentation-focused and internally consistent, with only a minor table clarity nit noted in the new page.
Review details
Suppressed comments (1)
Previously missed (1) — in code that hasn't changed since the last review.
chapters/ai-tools/posit-dev-ai-repos.qmd:37
- In the "Language" column, the value
noneisn’t a programming language and reads like a placeholder; using a neutral marker (e.g.,-) makes the table clearer and avoids implying a real language.
- Files reviewed: 4/4 changed files
- Comments generated: 0 new
- Review effort level: Lite
# Conflicts: # references.bib
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
There was a problem hiding this comment.
🟡 Changes recommended
references.bib has an invalid/malformed BibTeX entry (posit_dev_positron) that will break bibliography parsing/rendering.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Review details
- Files reviewed: 4/4 changed files
- Comments generated: 1
- Review effort level: Lite
The union resolver used for the sync dropped the blank line between every pre-existing entry and truncated one entry, which made pandoc reject the file and failed the build. The file is now rebuilt from the two parent versions: main's entries in main's order and formatting, then this branch's new entries appended, validated with pandoc. Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
There was a problem hiding this comment.
🟡 Changes recommended
The new Quarto content introduces a style-guide inconsistency in list formatting (and a small inline formatting inconsistency) that should be corrected before merging.
Once you've addressed the issues Copilot identified, you can request another Copilot review.
Review details
- Files reviewed: 4/4 changed files
- Comments generated: 2
- Review effort level: Lite
| The rest of the organization is Positron and its extensions, | ||
| Shiny for Python, `great-tables`, `pointblank`, `air`, `ggsql`, | ||
| Posit Team deployment tooling, container images, and workshop material. |
| | [`btw`](https://github.com/posit-dev/btw) [@posit_dev_btw] | Toolkit for giving LLMs context about an R session: copy context to the clipboard, chat in the IDE, or connect R to coding agents through `mcptools` | R | 135 | 2026-09-09 | Yes: the most direct fit for an R-centric workflow | | ||
| | [`mcp-repl`](https://github.com/posit-dev/mcp-repl) [@posit_dev_mcp_repl] | Sandboxed MCP server that keeps a persistent R or Python session alive across an agent's tool calls, with plots returned as images | Rust | 65 | 2026-07-27 | Yes, untried: replaces `Rscript -e` round trips in Claude Code | | ||
| | [`shiny-assistant`](https://github.com/posit-dev/shiny-assistant) [@posit_dev_shiny_assistant] | Source of the hosted Shiny Assistant chat app | Python | 58 | 2026-01-17 | No: an application, not a library | | ||
| | [`raghilda`](https://github.com/posit-dev/raghilda) [@posit_dev_raghilda] | Retrieval-augmented generation pipeline in Python, with DuckDB or ChromaDB storage | Python | 36 | 2026-09-02 | No: the R counterpart is `tidyverse/ragnar` | |
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
This comment has been minimized.
# Conflicts: # references.bib
|
Claude finished review — View run Review: PR #211 —
|
There was a problem hiding this comment.
🟢 Approval recommended
Changes are low-risk documentation updates with one minor style-guideline tweak suggested for readability.
Review details
Suppressed comments (1)
chapters/ai-tools/posit-dev-ai-repos.qmd:17
- The sentence listing the non-AI parts of the
posit-devorg contains 3+ technical items in comma-separated prose. Repo style guidance asks for a bullet list (with a blank line before it) for lists of 3+ items to improve readability in .qmd files (see.github/copilot-instructions.md:24-51).
The rest of the organization is Positron and its extensions,
Shiny for Python, `great-tables`, `pointblank`, `air`, `ggsql`,
Posit Team deployment tooling, container images, and workshop material.
- Files reviewed: 4/4 changed files
- Comments generated: 0 new
- Review effort level: Lite
|
Issue #106 asks for a review of the whole
posit-devGitHub organization.The lab already covers one of its repositories (
posit-dev/skills,@sec-ai-posit-skills),but Posit publishes a growing set of LLM packages, MCP servers, and agent tooling for R and Python
that a lab centered on R and Quarto should know about, and nothing in the site surveys them together.
This PR adds
chapters/ai-tools/posit-dev-ai-repos.qmd("Posit's AI Tooling on GitHub"),transcluded from
chapters/coding-agents.qmdright after the Posit skills section.The organization was enumerated with
gh repo list posit-devrather than by browsing,then filtered to the AI-related repositories;
the fragment tabulates them (purpose, language, stars, last push, verdict)
and closes with a short "Useful to us?" assessment for the lab's workflow.
Closes #106
🤖 Generated with Claude Code