Team Ops gives an agent two Python tools. One runs a structured team performance audit from role descriptions, OKRs and output data, ending in individual scorecards and an A, B or C stack rank. The other reads meeting transcripts and returns decisions, action items with owners and deadlines, open questions and follow-up meetings.
Both tools call an LLM through your own Anthropic or OpenAI key. Action items can optionally be pushed to HubSpot as tasks.
What You Get
- A performance report in Markdown or JSON with an executive summary, individual scorecards and a stack rank into A, B and C players
- Recommended actions per person (promote, coach, reassign) and org-level findings on overlap, bottlenecks and automation
- A structured meeting summary with decisions, action items, open questions, key quotes and follow-up meetings
- Confidence scores on each extracted item, with implicit commitments such as “I’ll look into it” flagged
- Optional HubSpot tasks created from the extracted action items
Common Use Cases
- Quarterly review of a team against its OKRs and KPIs
- Finding redundant roles and process bottlenecks before a reorganisation
- Turning a batch of recorded meetings into one follow-up list
- Checking that every commitment made in a call has an owner and a date
What Makes It Unique
Most meeting-notes tools stop at a summary. This one extracts implicit commitments with a confidence score and a source quote, and can push them to a CRM. The audit side pairs an LLM review of redundancy and automation with a locally computed score, so a ranking still comes out when no model key is set.
Under The Hood
The skill is a SKILL.md plus two Python scripts. The agent runs them from the shell. The scripts read the team data or transcript files you point them at, send the text to the Anthropic or OpenAI API with your key, and write a report file or print to the terminal. With a HubSpot token set, the extractor calls the HubSpot API to create tasks. The SKILL.md preamble also runs two telemetry helper scripts that log usage locally and ask once about opt-in remote telemetry. Without an LLM key the scripts return a notice, and the audit still computes its numeric scores locally.