Skip to content

RWA 2026.03.29

Choose a tag to compare

@github-actions github-actions released this 29 Mar 15:04
· 17 commits to master since this release

Script-first literature search workflow, Zotero attachment uploads, README Philosophy section, and infrastructure improvements.

Added

  • Script-first search workflow: draft_*_search and run_search_script tools added to all five database MCP servers (PubMed, OpenAlex, Semantic Scholar, Europe PMC, CrossRef). Formal literature searches now generate a standalone Python script for user review before execution, improving reproducibility and auditability.
  • Shared search script engine: generate_search_script, execute_search_script, and generate_and_run_script functions added to rwa_result_store shared library, powering the draft/run workflow across all servers.
  • Zotero attachment upload: upload_attachment tool added to zotero-server for uploading PDFs and other files to Zotero items via the Web API.
  • PDF upload utility scripts: scripts/_upload_pdfs.py and scripts/_upload_pdfs2.py for batch-uploading PDFs to Zotero library items.
  • Search results backfill utility: scripts/backfill_search_exports.py for regenerating Excel exports from existing search_results.db files.
  • Bug report issue template: .github/ISSUE_TEMPLATE/bug-report.yml for structured issue reporting.
  • README Philosophy section: Added project philosophy explaining RWA as a proof of concept, the case against platform lock-in, and the existing research tech stack, with a link to the LinkedIn launch post.
  • Search strategy template update: Added script-first workflow instructions to templates/systematic-review/search-strategy.qmd.
  • Agent instructions for script-first workflow: Updated systematic-reviewer and research-orchestrator agents with the mandatory draft-then-approve search protocol, including allowed direct MCP tool usage and DB-as-source-of-truth rules.
  • Global copilot instructions: Added Literature Search Protocol section enforcing the script-first, DB-as-source-of-truth workflow across all agents.

Changed

  • MCP server project path handling: Improved error messages and validation for project_path parameters across PubMed, OpenAlex, Semantic Scholar, Europe PMC, and CrossRef servers.
  • API documentation: Updated docs/api-setup-guide.md, docs/database-access.md, and docs/getting-started.md with clarified instructions.
  • Search runners: Improved search_runners.py in the shared library for better script execution handling.

Fixed

  • Chat parser: Handle non-string content in thinking blocks to prevent parsing errors in chat export.

Removed

  • Obsolete search_results.db files from sample project directories (data now lives in project-specific paths).
  • Stale prisma-flow.json from sample project (replaced by PRISMA tracker server workflow).