--- name: setup-callback description: This skill should be used when the user asks to "set up callback in this project", "start a job hunt here", "initialize callback", "bootstrap my job search", "set up my job search project", "init callback", or "configure callback for this project". Use it to scaffold project paths, onboard a resume, capture search preferences, and optionally set up California EDD/ledger tracking — in one pass. --- # Setup Callback ## North Star Land an interview by preparing an honest candidate profile and project structure before any job search or tailoring begins. ## Flow ### 1. Scaffold project On an existing project, load `.callback/config.json` first. Keep its `record_csv`, `ledger_db`, and `edd_xlsx` keys when present; remove them only after the later import succeeds with `skipped 0`. Write the config in the project root. Prompt for each path or accept defaults: ```json { "applications_dir": "./applications", "archive_dir": "./archive" } ``` Create the directories: `applications/`, `archive/`. Add `.callback/` to the project's `.gitignore` (append it; never replace the existing ignore rules) so automation memory and job-search details stay local. Re-running this step updates the config in place — no duplicates. ### 2. Onboard Ask for: - `resume_path` — PDF, DOCX, or TXT. - `accomplishments_path` — optional plain-text accomplishments file. Then: 1. Call `onboard_user(resume_path=..., accomplishments_path=...)`. 2. On success (`next_action: compile_profile`), call `compile_profile(session_id=...)` with its returned session ID. 3. If `accomplishments_path` is supplied, follow `onboard-profile`'s Scan → Plan → Confirm → Compile workflow to create the approved stories before reporting setup complete. 4. Report: registered label, detected sections, warnings, skill coverage gaps, next action. Never fabricate experience, skills, dates, metrics, or tools. ### 3. Capture preferences Ask these questions (skip any the user already answered): - **Location:** home city/state; remote preference. - **Work types:** one or more of `onsite_local`, `hybrid_local`, or `remote`. - **Comp target:** annual total comp (USD or omit). - **Target titles:** list of preferred job titles. - **Seniority bands + blockers:** bands you want (e.g. `["senior", "staff"]`); titles/levels to exclude. - **Target companies:** companies of interest. - **Core domains / skip domains:** domains to prioritize vs. skip. - **Referral companies:** companies where you have a contact (`name` + optional `note`). - **Scan sources:** structured sources, e.g. `[{"name": "Gmail alerts", "kind": "email", "instructions": "Search job-alert emails"}]`. `kind` is one of `email`, `web_search`, `careers_page`, or `job_board`. - **Lead recency (days):** how many days back to scan (default: 3). - **Sponsorship and work authorization:** whether sponsorship is needed, and the user's current authorization status. - **Actual years of experience:** the truthful total; use `yoe_gap_multiplier` `1.75` unless the user specifies another value. Then call `set_search_preferences(...)` with all answers. Note: this fully replaces stored prefs, so collect everything before calling. ### 4. Job history (optional) Job history lives in callback's global ledger. Build one `callback ledger import` command from only the legacy sources that actually exist: add `--archives ` for run archives, `--csv ` for an old record CSV, and `--contacts-db ` for an old `job_search_ledger.sqlite3`. Do not pass an absent source as a flag. After the import exits successfully with `skipped 0`, remove `record_csv`, `ledger_db`, and `edd_xlsx` from `.callback/config.json`. Keep those keys when the CLI lists any `skipped:` lines (malformed archive, bad date, ambiguous company/title, and so on) until the user fixes or explicitly discards every skip — a non-zero skip list means migration is incomplete even though the command exited 0. Keep them on a failed import too so the command can be retried. For an unemployment (EDD) report, use `/export-ledger`. ## Rules - No hardcoded personal values, absolute paths, or company/location defaults. - Paths live in `.callback/config.json`; preferences live in the profile via `set_search_preferences`. - Re-running init updates config and prefs in place. - Truthful evidence only — never fabricate or keyword-stuff.