[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"skill:01M756SPCT0TFXPDXPB1R2GEXJ":3,"skill-same:01M756SPCT0TFXPDXPB1R2GEXJ":53},{"skillNo":4,"name":5,"nameZh":6,"description":7,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":16,"categories":17,"keywords":18,"publishedAt":19,"linkOnly":20,"markdownContent":21,"markdownBody":22,"frontMatter":23,"refName":24,"commitSha":25,"skillDirectory":26,"forksCount":27,"repositoryHtmlUrl":28,"repositoryHomepageUrl":29,"ownerHtmlUrl":30,"ownerBlogUrl":31,"ownerDescription":6,"updatedAt":32,"install":33,"resources":46},"01M756SPCT0TFXPDXPB1R2GEXJ","webskrap",null,"Use when writing, debugging, documenting, or reviewing Python scraping, web search, and visual browser automation with WebSkrap. Not for ordinary HTTP clients or search APIs.","kacigaya\u002Fwebskrap","kacigaya","https:\u002F\u002Favatars.githubusercontent.com\u002Fu\u002F163902005?v=4","01M756SKCVA2SJG536A0MQK9Y5","Gaya KACI","01M756SKDXZE51613V4Z5AER3X",1,"Apache-2.0","https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fwebskrap\u002FSKILL.md",[],[],"2026-11-12 13:55:43",false,"---\nname: webskrap\ndescription: Use when writing, debugging, documenting, or reviewing Python scraping, web search, and visual browser automation with WebSkrap. Not for ordinary HTTP clients or search APIs.\n---\n\n# WebSkrap\n\nWebSkrap is an async Python scraping package that drives Chromium through its\nown Chrome DevTools Protocol (CDP) client, with no Playwright. The Python API,\n`webskrap` CLI, and `webskrap-mcp` server share core behavior.\n\nWebSkrap loads known URLs. To find URLs it loads DuckDuckGo or Bing's\nresults page in the same browser and parses the hits. It calls no search API\nand it does not offer Google.\n\n## Workflow\n\n1. Identify the affected public surface: Python, CLI, MCP, or more than one.\n2. Read its implementation and nearest tests. Trace calls into shared code\n   before editing.\n3. Change the shared implementation when behavior belongs to more than one\n   surface. Do not patch each wrapper independently.\n4. Update affected public contracts: models, serialization, errors, command\n   help, MCP schemas, guides, and tests.\n5. Run focused tests during development, then the full validation gate before\n   handoff when practical.\n\n## Route by task\n\n| Task | Implementation | Tests and contracts |\n| --- | --- | --- |\n| Fetching, sessions, links, or screenshots | `src\u002Fwebskrap\u002Fclient.py`, `src\u002Fwebskrap\u002Fmodels.py` | `tests\u002Ftest_client_unit.py`, `tests\u002Ftest_models.py` |\n| Search engines, redirect unwrapping, or results parsing | `src\u002Fwebskrap\u002Fsearch.py`, `src\u002Fwebskrap\u002Fmodels.py` | `tests\u002Ftest_search.py` and its fixtures under `tests\u002Ffixtures\u002Fsearch\u002F` |\n| Parsing | `src\u002Fwebskrap\u002Fparsing.py` | `tests\u002Ftest_parsing.py` |\n| Errors, hints, or exit statuses | `src\u002Fwebskrap\u002Ferrors.py`, `src\u002Fwebskrap\u002Fcli_output.py` | `tests\u002Ftest_errors.py`, CLI and MCP callers |\n| Main CLI commands | `src\u002Fwebskrap\u002Fcli.py` | `tests\u002Ftest_cli.py`, `tests\u002Ftest_diagnostics.py` |\n| Persistent browser state | `src\u002Fwebskrap\u002Fbrowser_session.py`, `src\u002Fwebskrap\u002Fpaths.py` | `tests\u002Ftest_browser_session_unit.py`, `tests\u002Ftest_paths.py` |\n| Interactive browser commands | `src\u002Fwebskrap\u002Fbrowser_cli.py`, `src\u002Fwebskrap\u002Fmcp_server.py` | `tests\u002Ftest_browser_cli.py`, `tests\u002Ftest_browser_integration.py`, `tests\u002Ftest_mcp.py` |\n| MCP tools, resources, or annotations | `src\u002Fwebskrap\u002Fmcp_server.py` | `tests\u002Ftest_mcp.py`, `src\u002Fwebskrap\u002Fguide.md` |\n| Profiles or stealth defaults | `src\u002Fwebskrap\u002Fprofiles.py`, `src\u002Fwebskrap\u002Fmodels.py`, `src\u002Fwebskrap\u002Fclient.py` | `tests\u002Ftest_profiles.py`, opt-in bot-detection tests when needed |\n| Cookie rejection | `src\u002Fwebskrap\u002Fconsent.py` | `tests\u002Ftest_consent.py`; `tests\u002Ftest_consent_live.py` only for selector rot |\n| Documentation and examples | `README.md`, `src\u002Fwebskrap\u002Fguide.md` | `tests\u002Ftest_docs.py`, current CLI help, public Python exports |\n\nSearch for every constructor, serializer, command, and schema that uses a\nchanged model or field. Fetch and search behavior cross Python, CLI, and MCP.\nPersistent browser behavior crosses the browser CLI and MCP. Engine markup\nlives only in `src\u002Fwebskrap\u002Fsearch.py`. When an engine changes its page,\nrefresh that engine's fixture and fix that one extractor.\n\n## Choose the public surface\n\n| Need | Python | CLI | MCP |\n| --- | --- | --- | --- |\n| Fetch one known URL | `client.fetch()` | `webskrap fetch` | `stealth_fetch` |\n| Find URLs for a query | `client.search()` | `webskrap search` | `search` |\n| Preserve cookies or storage | `client.session()` | `webskrap browser open` | `browser_open` |\n| See a rendered page or interact with canvas | `browser_session.view()` on a session page | `webskrap browser view` and `mouse` | `browser_view` and `browser_mouse` |\n| Click, fill, wait, or run a flow | `session.new_page()` and `human_click()` | `webskrap browser` | `browser_interact` and related tools |\n| Diagnose installation | Inspect raised error | `webskrap doctor` | `doctor` |\n\nUse a one-shot fetch for one page. Use a persistent browser only when state or\ninteraction must survive between operations. Do not repeatedly fetch a page to\ndrive a multi-step flow.\n\nRead only the reference needed for the task:\n\n- [Python API](references\u002Fpython-api.md): sessions, pages, profile modes, consent, links,\n  and fingerprint controls.\n- [CLI](references\u002Fcli.md): one-shot fetches, persistent browser commands, and\n  Linux ARM64 behavior.\n- [MCP](references\u002Fmcp.md): tool selection, confined paths, annotations, and\n  keeping tool results small.\n\n## Visual browser interaction\n\nUse `browser_view` (MCP) or `webskrap browser view --format json` (CLI) when\nlayout, images, canvas, or controls missing from the accessibility tree matter.\nMCP returns an image block directly. CLI saves a PNG; open its returned `path`\nwith the agent's image-viewing tool before choosing coordinates. A path or\nbase64 text alone does not give a model vision. The client and model must\nsupport image input.\n\nThe view covers the viewport at one PNG pixel per CSS pixel. Coordinates start\nat its top-left; use the original `width` and `height` if the image viewer\nresizes it. Do not add `scroll_x` or `scroll_y` to mouse coordinates. Full-page\nscreenshots are for inspection or export and do not map directly to the mouse.\n\nPrefer refs\u002Fselectors when they identify a control. Use `browser_mouse` \u002F CLI\n`mouse` for visual targets, then `browser_insert_text` \u002F CLI `insert-text` for\nliteral text in the focused control and `browser_press` \u002F CLI `press` for keys.\nWait for the expected change and take a fresh view after scrolling, navigation,\nor layout changes. Coordinate clicks hit whatever occupies that point.\nRun dependent actions sequentially within a session.\n\nThis is the browser's rendered page, available headless without Xvfb. It does\nnot show browser chrome, OS dialogs, or a desktop stream. One-shot fetch\nsessions are separate; open a persistent browser to drive a visual flow.\nRead the [CLI](references\u002Fcli.md) or [MCP](references\u002Fmcp.md) reference for examples.\n\n## Guardrails\n\n- Use public pages, local test servers, or targets the user is authorized to\n  access.\n- Do not add CAPTCHA solving, login-wall bypass, credential bypass, or\n  access-control circumvention.\n- Do not commit cookies, storage state, proxy credentials, or persistent browser\n  data such as `.webskrap\u002F`.\n- Preserve Chromium's sandbox by default. Any opt-out must remain explicit and\n  outside page-controlled input.\n- Keep MCP file paths confined to their configured roots. Reject absolute paths,\n  traversal, and symlink escapes.\n- Prefer public `webskrap` exports. Use private helpers only in tests.\n- Keep changes typed. Add focused tests for parsing, state transitions, output\n  contracts, and tool-safety behavior.\n\n## Public contract checklist\n\nWhen public behavior changes, check all applicable items:\n\n- Python models and exports\n- CLI arguments, help, JSON output, and exit status\n- MCP input schema, result shape, annotations, and server instructions\n- `README.md` and `src\u002Fwebskrap\u002Fguide.md`\n- unit tests for each affected surface\n\nUse the implementation as the source of truth. In particular, read\n`src\u002Fwebskrap\u002Fmodels.py` for result fields, `src\u002Fwebskrap\u002Ferrors.py` for error\ncodes and exits, and `src\u002Fwebskrap\u002Fdiagnostics.py` for environment reporting.\nDo not copy these catalogs into the skill.\n\n## Validation\n\nRun the nearest tests while editing. Before handoff, run the repository gate\nwhen practical:\n\n```bash\nuv run --extra dev pytest -q -m \"not browser and not live\"\nuv run --extra dev ruff check .\nuv run --extra dev ruff format --check .\nuvx --from pyright==1.1.411 pyright --pythonpath .venv\u002Fbin\u002Fpython src\u002Fwebskrap\nuv run --extra dev bandit -c pyproject.toml -q -r src\u002F\nuv run --extra dev --with pip-audit==2.9.0 pip-audit\nuv build\n```\n\nRun browser tests when browser behavior changes and the required browsers are\ninstalled.\n\nUse `WEBSKRAP_LIVE=1 pytest -q -m live` only when explicitly checking public\nthird-party behavior. Live tests are non-deterministic and do not replace the\ndefault suite.\n","# WebSkrap\n\nWebSkrap is an async Python scraping package that drives Chromium through its\nown Chrome DevTools Protocol (CDP) client, with no Playwright. The Python API,\n`webskrap` CLI, and `webskrap-mcp` server share core behavior.\n\nWebSkrap loads known URLs. To find URLs it loads DuckDuckGo or Bing's\nresults page in the same browser and parses the hits. It calls no search API\nand it does not offer Google.\n\n## Workflow\n\n1. Identify the affected public surface: Python, CLI, MCP, or more than one.\n2. Read its implementation and nearest tests. Trace calls into shared code\n   before editing.\n3. Change the shared implementation when behavior belongs to more than one\n   surface. Do not patch each wrapper independently.\n4. Update affected public contracts: models, serialization, errors, command\n   help, MCP schemas, guides, and tests.\n5. Run focused tests during development, then the full validation gate before\n   handoff when practical.\n\n## Route by task\n\n| Task | Implementation | Tests and contracts |\n| --- | --- | --- |\n| Fetching, sessions, links, or screenshots | `src\u002Fwebskrap\u002Fclient.py`, `src\u002Fwebskrap\u002Fmodels.py` | `tests\u002Ftest_client_unit.py`, `tests\u002Ftest_models.py` |\n| Search engines, redirect unwrapping, or results parsing | `src\u002Fwebskrap\u002Fsearch.py`, `src\u002Fwebskrap\u002Fmodels.py` | `tests\u002Ftest_search.py` and its fixtures under `tests\u002Ffixtures\u002Fsearch\u002F` |\n| Parsing | `src\u002Fwebskrap\u002Fparsing.py` | `tests\u002Ftest_parsing.py` |\n| Errors, hints, or exit statuses | `src\u002Fwebskrap\u002Ferrors.py`, `src\u002Fwebskrap\u002Fcli_output.py` | `tests\u002Ftest_errors.py`, CLI and MCP callers |\n| Main CLI commands | `src\u002Fwebskrap\u002Fcli.py` | `tests\u002Ftest_cli.py`, `tests\u002Ftest_diagnostics.py` |\n| Persistent browser state | `src\u002Fwebskrap\u002Fbrowser_session.py`, `src\u002Fwebskrap\u002Fpaths.py` | `tests\u002Ftest_browser_session_unit.py`, `tests\u002Ftest_paths.py` |\n| Interactive browser commands | `src\u002Fwebskrap\u002Fbrowser_cli.py`, `src\u002Fwebskrap\u002Fmcp_server.py` | `tests\u002Ftest_browser_cli.py`, `tests\u002Ftest_browser_integration.py`, `tests\u002Ftest_mcp.py` |\n| MCP tools, resources, or annotations | `src\u002Fwebskrap\u002Fmcp_server.py` | `tests\u002Ftest_mcp.py`, `src\u002Fwebskrap\u002Fguide.md` |\n| Profiles or stealth defaults | `src\u002Fwebskrap\u002Fprofiles.py`, `src\u002Fwebskrap\u002Fmodels.py`, `src\u002Fwebskrap\u002Fclient.py` | `tests\u002Ftest_profiles.py`, opt-in bot-detection tests when needed |\n| Cookie rejection | `src\u002Fwebskrap\u002Fconsent.py` | `tests\u002Ftest_consent.py`; `tests\u002Ftest_consent_live.py` only for selector rot |\n| Documentation and examples | `README.md`, `src\u002Fwebskrap\u002Fguide.md` | `tests\u002Ftest_docs.py`, current CLI help, public Python exports |\n\nSearch for every constructor, serializer, command, and schema that uses a\nchanged model or field. Fetch and search behavior cross Python, CLI, and MCP.\nPersistent browser behavior crosses the browser CLI and MCP. Engine markup\nlives only in `src\u002Fwebskrap\u002Fsearch.py`. When an engine changes its page,\nrefresh that engine's fixture and fix that one extractor.\n\n## Choose the public surface\n\n| Need | Python | CLI | MCP |\n| --- | --- | --- | --- |\n| Fetch one known URL | `client.fetch()` | `webskrap fetch` | `stealth_fetch` |\n| Find URLs for a query | `client.search()` | `webskrap search` | `search` |\n| Preserve cookies or storage | `client.session()` | `webskrap browser open` | `browser_open` |\n| See a rendered page or interact with canvas | `browser_session.view()` on a session page | `webskrap browser view` and `mouse` | `browser_view` and `browser_mouse` |\n| Click, fill, wait, or run a flow | `session.new_page()` and `human_click()` | `webskrap browser` | `browser_interact` and related tools |\n| Diagnose installation | Inspect raised error | `webskrap doctor` | `doctor` |\n\nUse a one-shot fetch for one page. Use a persistent browser only when state or\ninteraction must survive between operations. Do not repeatedly fetch a page to\ndrive a multi-step flow.\n\nRead only the reference needed for the task:\n\n- [Python API](references\u002Fpython-api.md): sessions, pages, profile modes, consent, links,\n  and fingerprint controls.\n- [CLI](references\u002Fcli.md): one-shot fetches, persistent browser commands, and\n  Linux ARM64 behavior.\n- [MCP](references\u002Fmcp.md): tool selection, confined paths, annotations, and\n  keeping tool results small.\n\n## Visual browser interaction\n\nUse `browser_view` (MCP) or `webskrap browser view --format json` (CLI) when\nlayout, images, canvas, or controls missing from the accessibility tree matter.\nMCP returns an image block directly. CLI saves a PNG; open its returned `path`\nwith the agent's image-viewing tool before choosing coordinates. A path or\nbase64 text alone does not give a model vision. The client and model must\nsupport image input.\n\nThe view covers the viewport at one PNG pixel per CSS pixel. Coordinates start\nat its top-left; use the original `width` and `height` if the image viewer\nresizes it. Do not add `scroll_x` or `scroll_y` to mouse coordinates. Full-page\nscreenshots are for inspection or export and do not map directly to the mouse.\n\nPrefer refs\u002Fselectors when they identify a control. Use `browser_mouse` \u002F CLI\n`mouse` for visual targets, then `browser_insert_text` \u002F CLI `insert-text` for\nliteral text in the focused control and `browser_press` \u002F CLI `press` for keys.\nWait for the expected change and take a fresh view after scrolling, navigation,\nor layout changes. Coordinate clicks hit whatever occupies that point.\nRun dependent actions sequentially within a session.\n\nThis is the browser's rendered page, available headless without Xvfb. It does\nnot show browser chrome, OS dialogs, or a desktop stream. One-shot fetch\nsessions are separate; open a persistent browser to drive a visual flow.\nRead the [CLI](references\u002Fcli.md) or [MCP](references\u002Fmcp.md) reference for examples.\n\n## Guardrails\n\n- Use public pages, local test servers, or targets the user is authorized to\n  access.\n- Do not add CAPTCHA solving, login-wall bypass, credential bypass, or\n  access-control circumvention.\n- Do not commit cookies, storage state, proxy credentials, or persistent browser\n  data such as `.webskrap\u002F`.\n- Preserve Chromium's sandbox by default. Any opt-out must remain explicit and\n  outside page-controlled input.\n- Keep MCP file paths confined to their configured roots. Reject absolute paths,\n  traversal, and symlink escapes.\n- Prefer public `webskrap` exports. Use private helpers only in tests.\n- Keep changes typed. Add focused tests for parsing, state transitions, output\n  contracts, and tool-safety behavior.\n\n## Public contract checklist\n\nWhen public behavior changes, check all applicable items:\n\n- Python models and exports\n- CLI arguments, help, JSON output, and exit status\n- MCP input schema, result shape, annotations, and server instructions\n- `README.md` and `src\u002Fwebskrap\u002Fguide.md`\n- unit tests for each affected surface\n\nUse the implementation as the source of truth. In particular, read\n`src\u002Fwebskrap\u002Fmodels.py` for result fields, `src\u002Fwebskrap\u002Ferrors.py` for error\ncodes and exits, and `src\u002Fwebskrap\u002Fdiagnostics.py` for environment reporting.\nDo not copy these catalogs into the skill.\n\n## Validation\n\nRun the nearest tests while editing. Before handoff, run the repository gate\nwhen practical:\n\n```bash\nuv run --extra dev pytest -q -m \"not browser and not live\"\nuv run --extra dev ruff check .\nuv run --extra dev ruff format --check .\nuvx --from pyright==1.1.411 pyright --pythonpath .venv\u002Fbin\u002Fpython src\u002Fwebskrap\nuv run --extra dev bandit -c pyproject.toml -q -r src\u002F\nuv run --extra dev --with pip-audit==2.9.0 pip-audit\nuv build\n```\n\nRun browser tests when browser behavior changes and the required browsers are\ninstalled.\n\nUse `WEBSKRAP_LIVE=1 pytest -q -m live` only when explicitly checking public\nthird-party behavior. Live tests are non-deterministic and do not replace the\ndefault suite.\n",{"name":5,"description":7},"main","914e24576ebe92e08d46c2867d065343b1b71fdc",".agents\u002Fskills\u002Fwebskrap",0,"https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap","https:\u002F\u002Fkacigaya.github.io\u002Fwebskrap\u002F","https:\u002F\u002Fgithub.com\u002Fkacigaya","https:\u002F\u002Fkacigaya.github.io","2026-11-15 02:23:26",{"source":8,"skillName":5,"commands":34,"prompt":44,"zipAvailable":45},[35,38,41],{"tool":36,"command":37},"npx","npx skills add kacigaya\u002Fwebskrap --skill webskrap",{"tool":39,"command":40},"bunx","bunx skills add kacigaya\u002Fwebskrap --skill webskrap",{"tool":42,"command":43},"pnpm","pnpm dlx skills add kacigaya\u002Fwebskrap --skill webskrap","安装技能，技能地址：https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fwebskrap\u002FSKILL.md",true,[47],{"path":48,"resourceType":49,"resourceTypeDesc":50,"sizeBytes":51,"sourceHtmlUrl":52},"references","OTHER","其他",19,"https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap\u002Fblob\u002Fmain\u002F.agents\u002Fskills\u002Fwebskrap\u002Freferences",[54,59],{"skillNo":55,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":56,"categories":57,"keywords":58,"publishedAt":19},"01M756SWCN06ASZ4Y3D7P8RWGB","https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap\u002Fblob\u002Fmain\u002F.codex\u002Fskills\u002Fwebskrap\u002FSKILL.md",[],[],{"skillNo":60,"name":5,"nameZh":6,"description":6,"descriptionZh":6,"repositoryFullName":8,"ownerLogin":9,"ownerAvatarUrl":10,"ownerAccountNo":11,"ownerDisplayName":12,"repositoryNo":13,"starsCount":14,"licenseSpdxId":15,"licenseFileUrl":6,"sourceHtmlUrl":61,"categories":62,"keywords":63,"publishedAt":19},"01M756ST53DBNQP59R0KNRANPT","https:\u002F\u002Fgithub.com\u002Fkacigaya\u002Fwebskrap\u002Fblob\u002Fmain\u002F.claude\u002Fskills\u002Fwebskrap\u002FSKILL.md",[],[]]