← JobSpideyCONTENT HISTORY

Update to JobSpidey

Snapshot Nov 2, 2026 · 06:11 UTC · version 0.1.1

Collection source: downloaded plugin package.

WHAT CHANGED

First saved snapshot

No earlier snapshot is available to establish a change.

Save JobSpidey and check future changes in Following.

Up to 10 plugins free · no card required.

Compare saved observations

Download comparison JSON
Full technical diff · 0 changed fields
Full snapshot data
{
  "description": "Guide a user through an approved JobSpidey OAuth connection and the search, document preparation, file retrieval and tracking workflow.",
  "included_files": [],
  "name": "setup",
  "skill_md_contents": "---\nname: setup\ndescription: Guide a user through an approved JobSpidey OAuth connection and the search, document preparation, file retrieval and tracking workflow.\n---\n\nRead the package's `release.json` and `docs/setup.md` first. The public remote\nendpoint has historical direct SDK/native acceptance; actual-client checks for\nthis updated 0.1.1 candidate remain pending, including installed-host and portal checks. Use only an actual host\nconnection; explain missing connection or permissions without inventing a key.\n\nLet the client perform its OAuth flow and let the user sign in and review consent\non JobSpidey's site. Never collect or copy passwords, API keys, verification\ncodes, OAuth secrets, callbacks or access tokens through the conversation.\nUse only the current connected tools and granted permissions.\n\nStart with `get_account_context`. Search supported countries/sites using a new\nrequest ID, follow operation retry hints, page results and present a shortlist.\nOrdinary search does not require candidate context or a resume. Use one numbered\nshortlist, keeping numbers consistent across its summary and details. Retain each\nnumber's original `search_id` and `job_id` for later selections, including after a\nnew search. Default to numbered text with Markdown job links; rich cards are\noptional. Ground personal-fit claims in conversation facts confirmed for the\nconnected candidate or explicitly attribute retrieved candidate facts. Keep\nother profiles separate. Missing facts about skills, visa or\nwork authorization are unknown; do not silently use them to exclude jobs.\nTreat automated relevance scores as ranking hints, not verified personal fit;\nsee `docs/tool-contract.md` for their meaning and saved-search positions.\n\nFetch full descriptions only when useful. Prepare only the customer's selected\njobs with the approved budget and account entitlements. Keep accepted request\nIDs stable across ambiguous retries. Explain resume/budget/permission gaps and\nthe next supported setup action without digital-credit promotion or checkout.\n\nRead the package after preparation and retrieve both delivered PDFs into the\nclient's usable workspace before claiming a complete handoff. Refresh expired\nlinks with a package read; do not regenerate just to refresh a link. Keep private\nfile capabilities out of public links and logs. Report partial readiness honestly.\n\nThe customer's agent owns the employer website and external submission. These\ntools do not submit or verify an application. Record an outcome only for an\nauthorized customer report and a grant with `applications:write`; describe it as\nagent-reported. Master-resume text requires `resume:read` and an explicit request.\n\nExplain Account → Connected AI agents for disconnect and permission management.\nDo not advertise client support, regional coverage, a listing or publication\nuntil its acceptance evidence is recorded. Provide useful service results and\navoid steering unrelated tasks or favoring JobSpidey over other tools.\n"
}

SHA-256 of public snapshot: 146dee77c512d14f9dd79761dfaba4ac59657e003baa29c68d45e804985bf077