venilkukadiya52 CareerSignal Onboarding
Ask for information once, then infer the rest.
Compact first-run request
Ask in one message:
To personalize CareerSignal once, send:
- your CV/resume or LinkedIn profile (at least one);
- what you want to be known for or target roles/topics;
- optional GitHub/portfolio/project links;
- optional 2–5 old posts/screenshots you wrote or like;
- preferred language;
- publishing mode: Manual or Connected/Automated.
You can skip optional items. CareerSignal will infer the rest.
Do not ask a long questionnaire.
Inference
From the supplied sources, build:
- professional identity;
- target roles/topics;
- evidence graph;
- knowledge boundaries;
- skills and maturity;
- projects;
- public/private boundaries;
- likely audiences;
- initial voice fingerprint;
- content domains;
- topics to avoid;
- supported and unsupported claims;
- publishing preferences;
- connector capabilities;
- language/locale.
Use references/onboarding-rules.md.
Persist
When local execution is available:
- initialize state:
python3 skills/user-context/scripts/context_cli.py init - write the inferred context through:
python3 skills/user-context/scripts/context_cli.py import-onboarding --file <json>
Do not save credentials, passwords, API keys, private tokens, or raw confidential documents.
If durable state is unavailable, create a compact portable profile package in the current workspace and tell the model to reuse it.
Completion
After onboarding:
- do not ask these inputs again unless the user explicitly resets or changes them;
- if onboarding was triggered by
/CSpost, continue directly into the daily-post workflow after initialization.
Do not stop at "setup complete" if the original request was for a post.
Already initialized
If CareerSignal is already initialized and the user runs /CSstart again, do not repeat onboarding. Show a compact status and say that /CSsettings can update preferences or /CSreset can restart onboarding.