R
A suite of four custom agent skills encoding PM domain knowledge — sprint velocity, billing compliance, agent orchestration, and token optimization. Real flows. Real outputs.
Parses raw ticket export CSVs into ranked team scorecards, week-on-week trend arrows, blocker age escalations, and auto-generated PM action items — all in under 2 minutes.
Multi-phase audit that queries live PSA/CRM data, validates rate card alignment, surfaces timecard approval gaps, checks purchase order alignment, and produces an action-first reconciliation dashboard.
Complete reference for designing, prompting, and debugging multi-step AI agents — covering mode selection, step-type rules, memory management, model routing, and data verification patterns.
Audits skills, prompts, and Claude Code workflows for token efficiency — applying a 4-question triage gate and 5 optimization levers. Reports realized vs theoretical savings with honest trade-off disclosure.
These skills encode domain logic that used to live only in people's heads — now they run on demand, in seconds, with the same output every time.
Talk to me ↗