Conversation
…nd support details test(docs): add test for broken repository-relative links in documentation
- Implemented the Skill Creator page to guide users through creating and validating skills. - Introduced UI components for defining task contracts, reviewing drafts, and accepting/exporting skills. - Added backend logic for skill validation and draft generation using AI. - Created tests for skill authoring and evidence librarian functionalities to ensure reliability. test: Add unit tests for skill authoring and evidence librarian - Developed comprehensive tests for skill authoring, including validation and draft generation. - Implemented tests for evidence planning, approval, and skill composition to ensure correct behavior. - Added behavioral tests for the Skill Creator page to validate user input requirements and export conditions.
feat: add skill authoring studio and Evidence Librarian
feat: add controls for active jobs
…nd support details test(docs): add test for broken repository-relative links in documentation
|
@LucaCappelletti94 I refreshed this PR onto current main without force-pushing and addressed the prior review by keeping link checking out of pytest. The final diff is now limited to README.md and docs/DESIGN.md, updated for Claude Code, Codex, OMP, all registered providers, and the current container architecture. Local lockfile, Ruff lint/format, and diff checks pass. Could you please re-review and approve the fork workflow run? GitHub currently marks CI as action_required pending maintainer approval. |
|
Up to standards ✅🟢 Issues
|
![codacy-production[bot]](https://avatars.githubusercontent.com/in/56611?s=60&v=4){kind=small}
There was a problem hiding this comment.
Pull Request Overview
The documentation updates successfully reflect the architectural shift to a PostgreSQL-backed, containerized multi-agent system. However, the PR is not yet ready for merging due to significant inconsistencies between the visual project structure and the actual commands provided for installation and execution. Most notably, the removal of the 'Makefile' and the 'job_manager' module from the README tree will lead to user confusion given their presence in bootstrap and build instructions.
Furthermore, the 'DESIGN.md' file introduces a 15-minute auto-continuation policy in Coinvestigate Mode which poses a safety and budget risk for users who expect a human-in-the-loop pause. Lastly, the README references several auxiliary documentation files that are not included in this commit, creating broken links. While Codacy results are up to standards, these alignment issues constitute an implementation gap relative to the goal of synchronizing documentation with the current state.
About this PR
- The documentation claims to align the current architecture but introduces references to 'docs/DEPLOYMENT.md' and 'docs/SECURITY_REVIEW.md' which are not included in the repository or this PR.
2 comments outside of the diff
README.md
line 188🟡 MEDIUM RISK
The bootstrap command references 'openscientist.job_manager', which is not listed in the new project structure (lines 104-115). If the module has been moved or replaced, the documentation for this command must be updated to prevent execution errors.
docs/DESIGN.md
line 137🟡 MEDIUM RISK
The 15-minute auto-continuation policy for Coinvestigate Mode presents a safety and budget risk. Users selecting this mode typically expect the agent to remain paused until explicit authorization is provided. Consider making this timeout configurable or requiring manual intervention to proceed.
Test suggestions
- Verify harness auto-selection logic correctly maps providers to harnesses (e.g., Anthropic to Claude Code)
- Verify execute_code tool correctly handles Rust and SPARQL execution within the executor container
- Validate PostgreSQL schema migrations for the knowledge_state (findings, hypotheses, literature)
- Confirm OMP harness functionality with non-vLLM/llama.cpp providers when explicitly selected
- Verify agent container egress policy correctly isolates jobs while allowing model provider access
Prompt proposal for missing tests
Consider implementing these tests if applicable:
1. Verify harness auto-selection logic correctly maps providers to harnesses (e.g., Anthropic to Claude Code)
2. Verify execute_code tool correctly handles Rust and SPARQL execution within the executor container
3. Validate PostgreSQL schema migrations for the knowledge_state (findings, hypotheses, literature)
4. Confirm OMP harness functionality with non-vLLM/llama.cpp providers when explicitly selected
5. Verify agent container egress policy correctly isolates jobs while allowing model provider access
TIP Improve review quality by adding custom instructions
TIP How was this review? Give us feedback
|
Review feedback is addressed in e4ac180:
The final diff against current upstream @LucaCappelletti94, could you please re-review to clear the earlier changes-requested state? A maintainer also needs to approve the fork workflow run: https://github.com/openscientist-io/openscientist/actions/runs/34217543741 |
2 participants
Summary
Updates project documentation to reflect the current OpenScientist implementation.
The contributor branch was synced with current upstream
mainwithout a force-push. The final diff against current upstreammainremains limited toREADME.mdanddocs/DESIGN.md.Validation
uv run --frozen pre-commit run --files README.md docs/DESIGN.mdgit diff --checkREADME.mdanddocs/DESIGN.md; all resolve.The full GitHub Actions workflow requires maintainer approval because this PR originates from a fork.