{"id":48464185,"url":"https://github.com/whittle/kosher","last_synced_at":"2026-04-07T04:02:04.475Z","repository":{"id":337950751,"uuid":"1155682076","full_name":"whittle/kosher","owner":"whittle","description":"Kosher updates BDD for the AI era. ","archived":false,"fork":false,"pushed_at":"2026-02-12T05:32:20.000Z","size":70,"stargazers_count":0,"open_issues_count":0,"forks_count":0,"subscribers_count":0,"default_branch":"main","last_synced_at":"2026-02-12T14:13:11.490Z","etag":null,"topics":["bdd","behavior-driven-development","inference","llm","mcp","mcp-server","slm","test-automation","testing","testing-framework","testing-tools","user-stories"],"latest_commit_sha":null,"homepage":"","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":null,"status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/whittle.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":null,"funding":null,"license":null,"code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null,"governance":null,"roadmap":null,"authors":null,"dei":null,"publiccode":null,"codemeta":null,"zenodo":null,"notice":null,"maintainers":null,"copyright":null,"agents":null,"dco":null,"cla":null}},"created_at":"2026-02-11T19:33:46.000Z","updated_at":"2026-02-12T05:32:23.000Z","dependencies_parsed_at":null,"dependency_job_id":null,"html_url":"https://github.com/whittle/kosher","commit_stats":null,"previous_names":["whittle/kosher"],"tags_count":null,"template":false,"template_full_name":null,"purl":"pkg:github/whittle/kosher","repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/whittle%2Fkosher","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/whittle%2Fkosher/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/whittle%2Fkosher/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/whittle%2Fkosher/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/whittle","download_url":"https://codeload.github.com/whittle/kosher/tar.gz/refs/heads/main","sbom_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/whittle%2Fkosher/sbom","scorecard":null,"host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":286080680,"owners_count":31499193,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2026-04-07T03:10:19.677Z","status":"ssl_error","status_checked_at":"2026-04-07T03:10:13.982Z","response_time":105,"last_error":"SSL_connect returned=1 errno=0 peeraddr=140.82.121.5:443 state=error: unexpected eof while reading","robots_txt_status":"success","robots_txt_updated_at":"2025-07-24T06:49:26.215Z","robots_txt_url":"https://github.com/robots.txt","online":false,"can_crawl_api":true,"host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["bdd","behavior-driven-development","inference","llm","mcp","mcp-server","slm","test-automation","testing","testing-framework","testing-tools","user-stories"],"created_at":"2026-04-07T04:02:03.770Z","updated_at":"2026-04-07T04:02:04.466Z","avatar_url":"https://github.com/whittle.png","language":"Python","readme":"# Kosher\n\nKosher is a Behavior-Driven Development (BDD) tool that reads Gherkin feature\nfiles and executes user stories against web applications using an AI inference\nengine and Playwright browser automation.\n\n## Requirements\n\nRequires ollama and playwright-mcp running locally. Tested using Python v3.14.\n\n## Getting Started\n\n```bash\n# 1. Install Ollama\ncurl -fsSL https://ollama.com/install.sh | sh\n\n# 2. Pull the model\nollama pull qwen2.5-coder:14b-instruct-q4_K_M\n\n# 3. Start Playwright MCP server (separate terminal)\nnpx @playwright/mcp@latest\n\n# 4. Clone and set up project\ngit clone \u003crepo\u003e\ncd kosher\npython -m venv venv\nsource venv/bin/activate\npip install -e .\n\n# 5. Run proof of concept\npython poc/main.py\n```\n## Reliability Summary: qwen2.5-coder:14b for Gherkin → Playwright\n\nSuccess Rate: ~90% on a 6-step login flow (20 runs)\n\nWhat Works Well:\n- Correctly interprets Given/When/Then semantics\n- Maps steps to appropriate browser tools (navigate, click, type, snapshot, wait_for)\n- Extracts element refs from snapshots and uses them correctly (most of the time)\n- Learns new patterns from system prompt updates (adopted browser_wait_for after instruction)\n\nFailure Modes:\n1. Placeholder refs - Sometimes outputs \u003cEMAIL-field-ref\u003e instead of actual ref like e5, suggesting it \"knows\" what to do but\ndoesn't execute properly\n2. Skipped tool execution - Occasionally outputs JSON + \"DONE\" in one response without waiting for tool results\n3. No native tool calling - Always outputs JSON in text content; requires parsing with parse_tool_call_from_text()\n4. Instruction drift - Multi-step instructions (navigate → snapshot → confirm) sometimes get partially followed\n\nImplications:\n- Viable for PoC and demos\n- Production use would need retry logic, validation, and possibly a more capable model\n- System prompt engineering is effective for teaching new patterns\n- The 10% failure rate is LLM variability, not fixable by prompt alone\n","funding_links":[],"categories":[],"sub_categories":[],"project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwhittle%2Fkosher","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fwhittle%2Fkosher","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fwhittle%2Fkosher/lists"}