Report — next steps for one opportunity

Role: Assistant — career / opportunity tracking (see docs/AGENT_ROLES.md).

Context to load:

• data/private/opportunities.yaml (private)
• Output skeleton: templates/opportunity_report_next_steps_one.md

Task:

1. Select the record by id: <OPPORTUNITY_ID> (or company + role if user specifies — resolve ambiguity).
2. Produce Snapshot + Next actions from next_action, next_action_date, notes, stage.
3. Populate Contacts table from contacts[]; if empty, write "None recorded."
4. Populate Check here (manual) table from links; if empty, write "None recorded."
5. Add Blockers / risks from notes and stage.
6. Add Suggested 24–72h moves aligned with the workflow style in docs/cursor/workflows.md (Workflow 2).
7. Assumptions / Risk.

Output: Markdown for reports/private/opportunity-next-steps-<id>-<YYYY-MM-DD>.md or paste-only.

Constraints:

• Full detail stays private; sanitize before any public reports/briefs/ excerpt.
• Do not fabricate. Every sentence in the report must be traceable to a field in the YAML record (stage, next_action, notes, history, contacts, links). Do not add general career advice, invented timelines, or assumed details not present in the data.
• Missing data → state it explicitly. If a section has no supporting data, write "No data recorded." Do not fill the gap with plausible-sounding content.
• Assumptions section is mandatory. Any interpretation beyond a direct field read (e.g. deriving a blocker from stage logic, suggesting a move based on recruiter note wording) must be listed there, clearly labeled as an inference.

My inputs (collect, confirm, or ask):

• If opportunity id (or company + role) is not in memory, chat, or unambiguous from opportunities.yaml: ask before generating the report.
• If you infer id from company + role: state the resolved id and ask the user to confirm or correct before final output.
• Opportunity id (or company + role): <...>