interviewstreet/hiring-agent
Every open issue, and what escalation-gate found for the ones it investigated. Updated 2026-10-11.
162open issues
71look like bug reports
192open pull requests
3investigated
1have a PR that fixes it and merges clean
| Issue | Kind | Title | Result | |
|---|---|---|---|---|
| ✅ | #476 | bug | Bug: identity month_map, unused fetch_profile prefix, and dead stream option | partly: #412, #477 → |
| ⬜ | #474 | other | chore: unused imports, debugger leak (import pdb), and dead config across codebase | — |
| ⬜ | #472 | bug | Bug: blog scoring pipeline is wired but never triggered (blog_data never passed to evaluator) | — |
| ⬜ | #470 | feature | Feature: optional evidence trace with source references and quote checks | — |
| ⬜ | #468 | bug | Bug: transform_achievements() drops the date emitted by the awards extractor | — |
| ⬜ | #459 | bug | Bug: GitHub enrichment is skipped when the resume writes the GitHub domain with capitals (GitHub.com/user) | — |
| ⬜ | #457 | bug | Gemini unusable with new API keys: retired models, hidden errors, quota retry loop | — |
| ⬜ | #452 | other | GitHub bio/repo description is unsanitized, injectable into the resume-scoring LLM prompt | — |
| ⬜ | #448 | bug | Bug: project highlights are extracted by the LLM but discarded in transform_projects() | — |
| ⬜ | #436 | bug | Minor: success response logged at ERROR, github.py uses pdf logger, 'top 5' message while selecting 7 | — |
| ⬜ | #435 | bug | Bug: provider does not handle null message content / finish_reason, crashing with 'NoneType' object has no attribute 'strip' | — |
| ⬜ | #434 | bug | Bug: GitHub rate-limit handling sleeps up to 1h but never retries the failed request | — |
| ⬜ | #433 | bug | Bug: GitHub project-selection exception fallback uses unfiltered repos (0 author commits) and drops commit counts | — |
| ✅ | #432 | bug | Bug: role.json min_final_score / max_final_score are loaded but never applied to the final score | partly: #292, #443 → |
| ⬜ | #431 | bug | Bug: resume/GitHub cache keyed only on PDF basename, so two different resume.pdf files share cached data | — |
| ⬜ | #430 | bug | Bug: CSV export rows shift into wrong columns when basics is missing (46 vs 31 columns under one header) | — |
| ⬜ | #429 | bug | Bug: project technologies are extracted but never included in the text sent to the evaluator | — |
| ⬜ | #428 | bug | Bug: one profile with url=null (or missing name) discards the entire basics section, silently disabling GitHub enrichment | — |
| ⬜ | #427 | bug | Bug: overall score is capped using the LLM-returned max instead of role.json weights (e.g. 105/165 instead of 60/100) | — |
| ⬜ | #426 | bug | Bug: prompt templates fail to load unless score.py is run from the repo root (template_dir is CWD-relative) | — |
| ✅ | #424 | bug | GitHub enrichment: project_type and contributor counts never reach the evaluator | merge #425 → |
| ⬜ | #409 | bug | Bug: "Open Source Contributions" resume section is silently dropped, but open_source is the highest-weighted scoring category (35/100) | — |
| ⬜ | #404 | bug | Fix: extracted Github project live demo URLs not being passed to LLM | — |
| ⬜ | #403 | other | providers.json num_ctx setting has no effect on Ollama and causes empty responses on CPU setups | — |
| ⬜ | #400 | bug | Bug: designed / multi-column resume layouts (Canva, etc.) silently drop whole sections during PDF to Markdown extraction | — |
| ⬜ | #398 | other | Docs: add reproducible uv setup using pinned Python 3.11.13 | — |
| ⬜ | #393 | feature | Feature: Score cover letters against a role's job description | — |
| ⬜ | #391 | feature | Add documentation and ease of setup for Anthropic endpoint | — |
| ⬜ | #388 | bug | fix: handle spaced date ranges and prevent endDate loss in parse_date_range | — |
| ⬜ | #386 | bug | fix: handle HTTP-date in Retry-After header during 429 retries | — |
| ⬜ | #383 | feature | Feature Request: Generate a revamped resume from the extracted data after scoring | — |
| ⬜ | #380 | feature | Add explicit Ollama support for gemma4:12b | — |
| ⬜ | #378 | bug | Bug: no Gemini model in providers.json is usable on a new API key (2.5-flash / 2.5-flash-lite return 404 "no longer available to new users") | — |
| ⬜ | #374 | bug | Bug: condensed "Also: ..." resume lines silently drop extra work/project entries | — |
| ⬜ | #373 | bug | Bug: shipped Anthropic provider config in providers.json is incompatible with real Claude models | — |
| ⬜ | #372 | bug | Bug: requirements.txt pins nonexistent versions of requests and python-dotenv, breaking fresh installs | — |
| ⬜ | #359 | bug | Docs: README still describes pre #298 provider set up, removed provider classes, and incorrect config | — |
| ⬜ | #356 | bug | Bug: .env's DEFAULT_MODEL silently ignored due to load_dotenv() import-order in config.py | — |
| ⬜ | #349 | bug | Gemini provider crashes with a Pydantic validation error on deductions.reasons | — |
| ⬜ | #345 | bug | Fenced JSON parsing fails after think blocks | — |
| ⬜ | #343 | feature | Feature: Add support for GitHub Copilot SDK as LLM Provider | — |
| ⬜ | #342 | other | GeminiProvider sends "system" messages as model turns instead of using system_instruction | — |
| ⬜ | #337 | other | Performance: GitHub API rate limit exhaustion and severe bottleneck during repository analysis | — |
| ⬜ | #335 | bug | Bug: extract_github_username keeps URL fragments and trailing punctuation, breaking GitHub enrichment | — |
| ⬜ | #331 | other | AI エージェント | — |
| ⬜ | #327 | other | github_project_selection prompt has contradictory rules on low-commit-count contributions | — |
| ⬜ | #321 | feature | feat: add llama.cpp and NVIDIA NIM provider support | — |
| ⬜ | #315 | other | PDF parsing issue and cannot access project active links | — |
| ⬜ | #312 | bug | Bug: extract_json_from_response fails to strip bare ``` code fences (no 'json' tag) | — |
| ⬜ | #310 | feature | feat: add Z.AI (GLM) provider support | — |
| ⬜ | #309 | other | Dribbble support | — |
| ⬜ | #308 | other | Create a golden dataset for evaluation of the agent’s performance | — |
| ⬜ | #300 | bug | Bug: Resume evaluation crashes with ValidationError when LLM returns negative deductions total | — |
| ⬜ | #296 | feature | feat: Add OpenCode API provider support | — |
| ⬜ | #295 | feature | [Feature] Add Z.ai (GLM) LLM provider | — |
| ⬜ | #288 | other | Profile username not extracted when network is already set (LinkedIn shows username: null) | — |
| ⬜ | #284 | bug | [Bug] Gemini provider aborts extraction on transient 504 DeadlineExceeded; only 429 is retried | — |
| ⬜ | #278 | bug | Ollama models fail to reliably extract resume sections (skills, projects, etc.) that work correctly with Gemini | — |
| ⬜ | #276 | other | Web UI to track resumes and get better insights | — |
| ⬜ | #273 | other | Security | Hidden PDF text poisons resume extraction and inflates hiring scores | — |
| ⬜ | #272 | feature | Feature Request: Add Docker Configuration for Local Development | — |
| ⬜ | #271 | other | Backend projects are penalized for missing live demos | — |
| ⬜ | #269 | other | GitHub project selector includes repositories below author commit threshold | — |
| ⬜ | #267 | bug | Logger INFO messages are incorrectly written to stderr | — |
| ⬜ | #265 | bug | Bug: Unicode emoji in print/log statements crash on Windows (cp1252) | — |
| ⬜ | #263 | feature | Feature: support Ollama Cloud as a third LLM backend | — |
| ⬜ | #262 | other | chore: Clean up unused standard library imports | — |
| ⬜ | #259 | feature | Feature: Add bullet point quality analysis to resume evaluation | — |
| ⬜ | #258 | bug | Bug: Invalid model in .env produces confusing Ollama 404 error instead of clear warning | — |
| ⬜ | #252 | feature | Feature: Skill Evidence Trace for Every Evaluation Score | — |
| ⬜ | #246 | feature | Feature: explain why each section got its score and how to improve it (rubric-based, no extra LLM call) | — |
| ⬜ | #240 | other | LLM hallucinates bonus points and open source contributions not present in resume or GitHub data | — |
| ⬜ | #228 | other | Open Source score ignores merged PRs to external repos (classifier only inspects own repos) | — |
| ⬜ | #226 | other | Extraction aborts and discards all sections when one optional section returns empty | — |
| ⬜ | #223 | other | Section extraction aborts silently when an LLM returns an empty object {} for a section | — |
| ⬜ | #220 | feature | [Improvement] Update Gemini model registry in prompt.py — missing some Gemini 3.x and Gemma 4 models | — |
| ⬜ | #217 | bug | Repo contributor count is wrong: /contributors response is paginated but only first page is read | — |
| ⬜ | #216 | bug | GitHub stats columns in exported CSV are always 0/empty (wrong dict level) | — |
| ⬜ | #209 | bug | Bug: Blog post data is excluded from evaluation due to missing fetch logic in main() | — |
| ⬜ | #208 | feature | feat: add OpenAI API provider support | — |
| ⬜ | #204 | feature | Add Claude/Gemini CLI and Anthropic Claude API as LLM backends | — |
| ⬜ | #202 | bug | Bug: Section extraction silently returns empty {} on small Ollama models due to JSON-schema format constraint | — |
| ⬜ | #196 | bug | [Bug]:Application Crashes When LLM Evaluation Fails | — |
| ⬜ | #194 | bug | Bug: Newer Gemini models (3.5-flash, 3.1-flash-lite) silently fall back to Ollama provider | — |
| ⬜ | #191 | feature | Feature: Support Coding Profile Enrichment (LeetCode, Codeforces, etc.) | — |
| ⬜ | #189 | feature | [feature] Add support to configure role and organization through .env parameters | — |
| ⬜ | #188 | feature | [improvement] add latest gemini models to ModelProviderMapping | — |
| ⬜ | #187 | feature | [improvement] use latest patch version for python 3.11 | — |
| ⬜ | #186 | bug | [Bug] Agent fails to complete response when Gemini Free Tier RPM limit is exceeded | — |
| ⬜ | #185 | bug | [Bug] Cache being saved and reused in case of error from model provider | — |
| ⬜ | #183 | feature | Add Profile Authenticity / Credibility Assessment | — |
| ⬜ | #175 | feature | Proposal: Extend Evaluation Beyond GitHub (LeetCode, HackerRank, CodeForces, CodeChef, etc) | — |
| ⬜ | #172 | feature | Feature: Add LeetCode Enrichment and LLM-Based Problem-Solving Evaluation | — |
| ⬜ | #166 | feature | [Feature Request] JD-Augmented Scoring and Multi-Candidate Batch Processing | — |
| ⬜ | #162 | bug | Bug: Project details of forked repository are not accounted for and skipped from all_repositories because of false positive of forks_count<5 | — |
| ⬜ | #160 | feature | Add comprehensive test suite for improved code reliability | — |
| ⬜ | #158 | feature | feat: Added role fit scoring | — |
| ⬜ | #156 | feature | Feat: Support for Multiple LLM Providers | — |
| ⬜ | #155 | bug | Bug: Forked repository stats are sourced from the user's fork instead of the upstream parent repository | — |
| ⬜ | #153 | feature | Feat: Add gitlab project extraction | — |
| ⬜ | #152 | feature | Feature Request: Extract clickable URLs from PDF text | — |
| ⬜ | #149 | bug | Fix fork contribution detection in GitHub analysis | — |
| ⬜ | #144 | feature | Add batch processing support for multiple resume evaluation | — |
| ⬜ | #139 | feature | Feature Request: Add Structured Output Support for Gemini API | — |
| ⬜ | #137 | other | Improve PDF Parser Robustness for Diverse Resume Layouts and Styles | — |
| ⬜ | #135 | other | Resume data showing as null due to stale cache files | — |
| ⬜ | #131 | bug | Issue: Inaccurate Open Source Contribution Detection | — |
| ⬜ | #126 | feature | Feature Request: Add support for .docx resume files | — |
| ⬜ | #119 | bug | Bug in fetch_repo_contributors function – unreachable code and incorrect return | — |
| ⬜ | #118 | other | Performance Bottleneck : High Latency and Slow Performance | — |
| ⬜ | #116 | other | 'charmap' codec can't encode characters when caching GitHub data | — |
| ⬜ | #114 | feature | [Feature] Profiler for testing performance | — |
| ⬜ | #112 | feature | [Feature] API endpoint for testing from different devices | — |
| ⬜ | #110 | other | [Security]: Data Validation Bypass – Missing Input Sanitization in URL Processing | — |
| ⬜ | #108 | bug | [Bug]: Race condition in concurrent cache file access | — |
| ⬜ | #104 | bug | Bug: Missing error handling for None response in evaluator.py causing crashes | — |
| ⬜ | #102 | bug | [Bug] : Memory leak in PDF processing due to unclosed PyMuPDF document objects | — |
| ⬜ | #100 | other | Invalid or incomplete PDFs being sent to LLM for evaluation | — |
| ⬜ | #98 | bug | [Bug] Variance in Resume Scores | — |
| ⬜ | #95 | bug | [Bug] "achievements" from markdown were not picked up by LLM | — |
| ⬜ | #94 | bug | [Bug]: pymupdf_rag.py was missing an entry function for development | — |
| ⬜ | #93 | other | [CI/CD] feat: Add CI/CD workflow for Code Quality | — |
| ⬜ | #92 | other | [Infra] Add Continuous Integration and Pre-Commit Setup | — |
| ⬜ | #90 | other | code-formatting: import statements are not PEP 8 compliant | — |
| ⬜ | #86 | bug | fix: (NameError) Undefined 'response' variable in github.py | — |
| ⬜ | #85 | feature | Feature Request: Expanded Social Profile Enrichment for Candidate Evaluation (adding twitter, blogs on mediums etc.) | — |
| ⬜ | #82 | bug | Bug: Extraction and evaluation of open source contributions is broken | — |
| ⬜ | #81 | feature | Feature : Add Multi-format Resume Support | — |
| ⬜ | #80 | feature | Feature Request: Resume Comparison System | — |
| ⬜ | #78 | feature | Feat : Adding Support for GROQ provider | — |
| ⬜ | #77 | bug | Bug: LLM_PROVIDER IS not used at all | — |
| ⬜ | #76 | feature | Feature Request: Add Scholarship/Mentorship Scoring as a Separate Category | — |
| ⬜ | #73 | feature | ✨Feature: Add configuration file support for customizable settings | — |
| ⬜ | #71 | feature | Add input validation for PDF files in extract_text_from_pdf function | — |
| ⬜ | #68 | bug | Issue: Missing Error Handling for JSON Parsing in extract_json_from_response | — |
| ⬜ | #65 | feature | Add extraction support for 'certificates' and 'volunteer experience' resume sections | — |
| ⬜ | #64 | other | Blog Data Evaluation Feature Incomplete: Missing Blog Content Fetching and Resume URL Extraction | — |
| ⬜ | #61 | other | GitHub Profile Data Not Exporting to CSV | — |
| ⬜ | #57 | other | Enhance Output Formatting and Readability | — |
| ⬜ | #56 | other | Minor: Repetitive use of Description key in project object formation while fetching GitHub repos | — |
| ⬜ | #52 | feature | Enhancement: Exclude bot accounts from contributor count before project classification | — |
| ⬜ | #46 | bug | Code after return statement never runs in `fetch_repo_contributors` | — |
| ⬜ | #44 | bug | Fix Undefined Variable in GitHub API Error Handling | — |
| ⬜ | #43 | other | Improve skills extraction to prioritize relevant skills instead of first 10 alphabetical skills | — |
| ⬜ | #40 | other | Optimize Resume Parsing via Sectional Extraction and Context Snippets | — |
| ⬜ | #37 | bug | Issue : Reference to Undefined Variable "response" in fetch_all_github_repos() | — |
| ⬜ | #36 | bug | UnicodeEncodeError and JSONDecodeError in score.py when running on Windows | — |
| ⬜ | #35 | bug | Issue: Non-deterministic scoring with Gemini Pro models even at temperature 0 | — |
| ⬜ | #33 | bug | Robust cache file handling for encoding errors. | — |
| ⬜ | #32 | feature | feat: Add support for certifications and publications sections in resume parsing | — |
| ⬜ | #31 | other | Could we add other Open source programs as well | — |
| ⬜ | #30 | feature | Enhancement: Advanced GitHub Analytics for Comprehensive Developer Assessment | — |
| ⬜ | #27 | other | Migrate to Async/Await for LLM API Calls | — |
| ⬜ | #24 | bug | Bug: Negative deduction total can crash resume evaluation with Pydantic validation error | — |
| ⬜ | #20 | other | Work section in resumecache_<resume_name>.json is always null | — |
| ⬜ | #18 | other | Inconsistent capturing of Work experience and what model is being used | — |
| ⬜ | #16 | other | Basics section in resumecache_<resume_name>.json is always null | — |
| ⬜ | #15 | feature | feat: Automate template discovery in TemplateManager | — |
| ⬜ | #14 | bug | CSV export missing GitHub data due to incorrect data structure access | — |
| ⬜ | #12 | feature | Enhancement: Improve Open Source Detection with PR Tracking and License-Based Classification | — |
| ⬜ | #9 | bug | Fix GitHub API reliability: encoding issues, retry logic, rate limit handling, and undefined variables | — |
| ⬜ | #8 | other | Update deprecated `google-generativeai` dependency | — |
⬜ not investigated · ✅ citations verified · ⚠️ a citation failed twice, needs a human · ✗ timed out or errored. "Kind" is a guess from the title. Each result is judged from the diffs, not run; open a packet for the evidence and its limits.