assafelovic / gpt-researcher
Code health
This codebase scores 7.2 out of 10 for code health, which we rate good. It also scores maintainability 8.3 and static performance 9.8 out of 10. The three are scored separately and never blended into one number. Risk is concentrated, as it usually is: 7 of 614 files are git hotspots, and they average 2.0 — which is where the fixes pay off most.
Do next
4 things worth doing this quarter.
Worth planning
Break up create_chat_completion_with_tools (CCN 24, 109 lines)
gpt_researcher/utils/tools.py:158
create_chat_completion_with_tools nests its control flow deeply; 1 file imports it, changed 1 time in 90 days.
- gain
- up to +2.5 health on this file(inferred)
- finding
- Nested complexity
- size
- CCN 24, 109 lines, nests 8 deep
- severity
- critical
Medium effort, worth a checkFlatten the nesting in parse_research_results_response
gpt_researcher/skills/deep_research.py:152
parse_research_results_response nests its control flow deeply, and part of it is duplicated, so splitting it also removes a copy; 4 files import it, changed 5 times in 90 days.
- gain
- up to +1.6 health on this file(inferred)
- finding
- Nested complexity
- size
- CCN 25, 53 lines, nests 5 deep
- severity
- high
Medium effort, worth a checkDelete 20 unused symbols and files (976 lines)
Nothing in the graph reaches them, across 15 files; every reader and agent pays to skip them.
- symbols
- 20
- lines
- 976
- safe to delete
- Yes(inferred)
Medium effortBreak up from_provider (CCN 41, 170 lines)
gpt_researcher/llm_provider/generic/base.py:206
from_provider is long, branchy and widely depended on; 10 files import it, changed 6 times in 90 days.
- gain
- up to +1.0 health on this file(inferred)
- finding
- Brain method
- size
- CCN 41, 170 lines
- severity
- critical
Medium effort, worth a check
Improve what Repowise can see
Add a test coverage report
Without one, Repowise cannot tell tested code from untested code, so every test-related action says “unknown”.
- line coverage
- Unknown, no report
Small effort
- Lines of code
- 63.7K
Recent activity
All commitsCommits
- ec3cdb0edocs(skill): fix drift in .claude reference docs2mo ago
- 8d57e690chore(multi_agents): nest AG2 variant under multi_agents/ag22mo ago
- bb3ec9ebAdd GetXAPI retriever for X/Twitter search2mo ago
- a99431caAdd Claude 4.x family to NO_SUPPORT_TEMPERATURE_MODELS2mo ago
- 17fe3f4efeat: upgrade MiniMax default model to M32mo ago
- 365bc359fix: normalize curate_sources() return type before assigning to researcher.context2mo ago
Decisions
- Handle List[dict] context from CURATE_SOURCES=True in deep researchproposed · 1mo ago
- Concatenate all pages in PyMuPDFScraperproposed · 1mo ago
- Fetch arXiv via arxiv Client API resultsproposed · 1mo ago
- Fix MCP transport auth by forwarding connection_headersproposed · 1mo ago
- Truncate document content to avoid embedding token limit errorsproposed · 1mo ago
- Coerce none/null/empty for Optional env fieldsproposed · 1mo ago
Everything we found
1,044 open
Every detector's open count by area, each with its worst example. An inventory, not a to-do list: the actions above are the part worth doing.
- worst item is critical severity. Code health659Untested hotspot · gpt_researcher/skills/researcher.py35 findings · 9.6 deducted
- worst item is medium severity. Refactoring64<anonymous>+1.5 health, S effort · frontend/nextjs/public/workbox-f1770938.js
- worst item is medium severity. Decisions23gpt_researcher/skills/deep_research.py16 awaiting review · 7 hotspots ungoverned
- worst item is medium severity. Ownership258gpt_researcher/retrievers/utils.py89% single-owner concentration, 4 commits in 90d
- worst item is low severity. Dead code40evals/hallucination_eval/run_eval.pyUnreachable symbol (90 lines)
Composition
Open the graphExplore this codebase
- Docs404 pages across 10 modules
- ChatAsk this codebase a question and get an answer with its sources.
- Files614 files with per-file docs, health and history
- ArchitectureDependency graph, layers, and 23 entry points
- Code healthPer-file scores, 731 open findings, coverage and refactoring targets
- Knowledge graphEntities, communities, and the paths between them
- Change couplingFiles that keep changing together, mined from commit history
- CommitsChange-risk ranked history with agent provenance
- ContributorsBus factor, per-file maintainers, and the human/agent split
- StatsSize class, origin, lifetime churn, rhythm and records
- CostsWhat indexing this snapshot cost, by model and by run.