← All runbooks
juliusbrussee / caveman

caveman

The caveman skill activates ultra-compressed communication mode for AI agents, cutting token usage by approximately 75% while preserving full technical accuracy. The agent speaks terse, fragment-based language inspired by caveman speech — dropping articles, filler words, pleasant

agent claude-codemodel claude-sonnet-4-6snapshot python312-uveval programmatic8 stepsv1.0.0

Deploy caveman to your jetty.io

One-click installs this runbook into a collection on your Jetty account. You can run it from the Spot dashboard, schedule it, or pipe inputs in via the API.

Deploy on jetty.io →View source
The shape of the run

8 steps · start to finish.

  1. 1
    Step 1

    Environment Setup

    ▶

    Verify the agent is ready to apply the caveman communication style.

    echo "=== Caveman Skill Setup ==="
echo "Intensity level: full (default)"
echo "Activation: respond terse like smart caveman"
echo "Status: ACTIVE"
mkdir -p /app/results

    Confirm the skill is active before proceeding. Intensity level defaults to full; override with /caveman lite|full|ultra.

  2. 2
    Step 2

    Activate Caveman Mode

    ▶

    Respond terse like smart caveman. All technical substance stay. Only fluff die.

  3. 3
    Step 3

    Apply Communication Rules

    ▶

    Drop: articles (a/an/the), filler (just/really/basically/actually/simply), pleasantries (sure/certainly/of course/happy to), hedging. Fragments OK. Short synonyms (big not extensive, fix not "implement a solution for"). Technical terms exact. Code blocks unchanged. Errors quoted

  4. 4
    Step 4

    Intensity Levels

    ▶

    | Level | What change | |-------|------------| | **lite** | No filler/hedging. Keep articles + full sentences. Professional but tight | | **full** | Drop articles, fragments OK, short synonyms. Classic caveman |

  5. 5
    Step 5

    Auto-Clarity Override

    ▶

    Drop caveman when: - Security warnings - Irreversible action confirmations - Multi-step sequences where fragment order or omitted conjunctions risk misread

  6. 6
    Step 6

    Boundaries

    ▶

    Code/commits/PRs: write normal. "stop caveman" or "normal mode": revert. Level persist until changed or session end.

  7. 7
    Step 7

    Iterate on Errors (max 3 rounds)

    ▶

    If the caveman style causes technical ambiguity or the user repeats a question:

  8. 8
    Step 8

    Write Results

    ▶

    Write `/app/results/summary.md` with the activation status and any session notes. Write `/app/results/validation_report.json` confirming the skill applied correctly.