| name | job-researcher |
| description | STEP 1 of the application pipeline. Receives raw job input (files in raw_inputs/job/, pasted text, screenshot, or URL) and produces active_application/job_description.md enriched with company intelligence. Trigger: user asks to process a new job posting. MUST run before application-advisor. Do NOT skip this step even if the user asks to go straight to the advisor. |
Job Researcher
Turns raw job input into an objective, structured active_application/job_description.md by combining the posted JD with verifiable company intelligence.
Inputs
- Files in
raw_inputs/job/ (supported formats: pdf, txt, md; .md preferred)
- Pasted text, screenshot, or URL
- Company name (extract from JD if not stated explicitly)
Workflow
- Accept either chat input (paste/screenshot/URL) or files in
raw_inputs/job/.
- Extract the core job details from the user's input (role, company, location, requirements).
- Research the company using web search. Target sources in this order:
- Company website (About, Mission, Product/Services pages)
- LinkedIn company page (headcount, recent posts, culture)
- Recent news (funding, launches, press coverage from the last 12 months)
- Glassdoor or similar (culture signals, interview process if available)
- Synthesise everything into
active_application/job_description.md using the output format below.
- Ask the user: "Anything missing or worth adding before we move to the Advisor?"
Output format for active_application/job_description.md
# Job Description
## Role & Basics
- **Title**:
- **Company**:
- **Location / Remote**:
- **Employment type**:
## Responsibilities
[Bullet list from the posting]
## Requirements
### Must-have
### Nice-to-have
## Company Intelligence
### What they do
[1–2 sentence summary of the product/service]
### Mission & Values
[From website or LinkedIn]
### Culture signals
[Team size, work style, perks, tone of their communications]
### Recent news & context
[Funding rounds, product launches, notable hires — last 12 months]
### LinkedIn insights
[Headcount, growth trend, key people in the team/department]
## Hiring Signals (objective)
- [Exact terms, priorities, and repeated keywords from JD + company materials]
- [Operational constraints: language, location, tool stack, reporting expectations]
Rules
- Always save to
active_application/job_description.md — this is the single source of truth for all downstream agents.
- Prefer specificity over generality — the more concrete the intelligence, the better the cover letter.
- Preserve the exact section names in the output format so downstream skills can rely on them, especially
## Company Intelligence and ## Hiring Signals (objective).
- Stay objective. Do not evaluate candidate fit, recommend narrative angles, or suggest what the user should emphasise.
- Do not write or modify
active_application/application_brief.md or active_application/final_draft.md.
