socialscience.ai / workshop / day-1

§ i · Monday, 1 June 2026

Foundations & first real workflow

Deliverable
Working Claude Code env + your own global CLAUDE.md + a cleaned, aggregated dataset with assertion checks.

Slides

Open deck in presentation mode ↗

Schedule

10:00 AM – 3:00 PM ET · 255 min content · 45 min breaks + lunch

  1. 10:00
    Welcome + harness vs. model framing + landscape (Claude Code, Cursor, Codex; CLAUDE.md vs. agents.md)Lecture
    20 min
  2. 10:20
    Setup + first conversation: confirm everyone is talking to Claude CodeWorkshop
    30 min
  3. 10:50
    BreakBreak
    15 min
  4. 11:05
    Permission model + slash commands tour + --dangerously-skip-permissionsLecture
    15 min
  5. 11:20
    Configure permissions + try slash commands hands-on; ask Claude Code to install R or Python and the workshop packages (first real 'use Claude Code to do work' task)Workshop
    25 min
  6. 11:45
    Plan / Execute / Clear loop + named keystrokes (Shift+Tab, ESC, voice) + lightweight assertionsLecture
    20 min
  7. 12:05
    What plays well with agentic tools (formats; R/Python vs. SPSS/Stata)Lecture
    20 min
  8. 12:25
    Data profiling on the workshop dataWorkshop
    20 min
  9. 12:45
    LunchBreak
    30 min
  10. 1:15
    Data cleaning + aggregation with assertion checks (begins with 5-min walkthrough of the provided workshop project CLAUDE.md)Workshop
    30 min
  11. 1:45
    Reading Claude's output; debugging when Claude errors; 'ask Claude for everything' habitLecture
    15 min
  12. 2:00
    Continue cleaning + first own-data stepWorkshop
    20 min
  13. 2:20
    CLAUDE.md basics + interview-trick demoLecture
    15 min
  14. 2:35
    Interview trick: generate the smallest global CLAUDE.md that changes behavior; verify with the 'hide a secret' testWorkshop
    20 min
  15. 2:55
    Debrief + Day 2 previewDiscussion
    5 min

Materials

← All days