Precision Transcript Extraction
Transform messy meeting transcripts into high-signal action plans. This skill solves the problem of "meeting amnesia" by distinguishing between firm commitments and tentative ideas that often get lost in the chatter. Unlike basic summarizers that provide vague bullet points, this tool rigorously classifies tasks based on speaker intent and owner attribution.
What it does
- Intelligent Classification: Categorizes items as Firm (committed), Tentative (to be revisited), or Needs-Confirmation.
- Owner Attribution: Automatically maps tasks to specific speakers based on transcript labels.
- Safety Guardrails: Prevents "artificial certainty" by marking unattributed decisions for human review rather than guessing owners.
- Structural Integrity: Outputs clean Markdown that identifies exactly what was promised and by whom.
Why use this skill?
Prompting an AI yourself often leads to "hallucinated commitments" where the AI turns a casual suggestion into a hard task. This skill uses a specialized extraction workflow that respects the nuance of spoken language. It is optimized for developers and project managers who need reliable, audit-ready to-do lists from internal syncs or client calls.