The method behind the Rigor Pack

    The Skill
    Distillation Kit

    Skill distillation is the community's name for having a stronger model write the skills your daily model follows. Nothing about the model is distilled: what transfers is written procedure. This kit packages that procedure as three ready prompts, a blind-test task template, a grader rubric, and the protocol that separates real skills from placebo.

    The method, in four steps

    This is the exact loop that produced the six Rigor Pack skills, written with Claude Fable 5 during its July 2026 included-access window and blind-tested on Opus 4.8. The steps are public. The kit is the packaged execution of them, and it works with any stronger model you can reach.

    Pick a discipline gap, not a task.
    A skill that says write my unit tests is a prompt. A skill that says no edits until a written plan exists is a discipline worth keeping.
    Have the smarter model write it.
    In its own voice, about its own behavior, with frontmatter and rules concrete enough to be checkable.
    Blind-test it, with and without.
    Run the same task twice on your daily model. A separate grader judges the pair without knowing which run used the skill.
    Keep the losses.
    A results page that only shows wins is marketing. The losses are what make the wins believable.
    The numbers this method produced
    12 wins, 0 losses, 2 ties
    across 14 blind gradings on Opus 4.8
    2 first versions lost
    diagnosed, rewritten, re-tested on held-out tasks. Both runs published.
    Roughly 7% token overhead
    the measured cost of loading a skill, per task

    The six skills came out of this exact loop, written with Claude Fable 5 and graded blind on Opus 4.8. Every task, rubric, and run is published. Not affiliated with or endorsed by Anthropic.

    SKILL.md is an open standard. The same skill folder runs in Claude Code, Codex CLI, and Gemini CLI. The kit includes the per-tool install table.

    Get the kit

    One email with your access link to the full kit page.

    • Three ready skill-writing prompts: the gap interview, the skill author, the revision diagnosis
    • A blind-test task template with planted-trap slots
    • The grader rubric template and the run protocol
    • The per-tool install table and pointers to the six tested skills

    Free, delivered by email. Unsubscribe anytime.

    Just want the six finished skills? They stay free with no signup at the Rigor Pack.

    What is in the kit, and what is already free

    The gate is convenience, not exclusivity. Here is the honest split.

    In the kit

    • The three skill-writing prompts, ready to paste
    • The blind-test task template with planted-trap slots
    • The grader rubric template
    • The iteration protocol, with the worked example of a losing skill fixed

    Already free, no email needed

    Iwo Szapar
    Created By

    Iwo Szapar

    • •Co-created AI Maturity Index with Harvard
    • •Trained 25,000+ professionals on AI
    • •Ships blind-tested skills with every run published, including the losses

    Stop guessing whether your skills work

    Get the prompts and the test templates. Write skills with receipts.

    Free, delivered by email.

    Skill Distillation Kit and Second Brain 2.0

    The kit gives your model discipline you can prove. What it cannot give is memory: skills carry procedure, not context, and everything your agent learns about your work still evaporates when the session ends.

    The kit provides:

    • • Skill-writing prompts
    • • Blind-test templates
    • • The iteration protocol
    • • Proof discipline for any agent

    Second Brain 2.0 adds:

    • • Memory layer (SQL backbone)
    • • 30+ specialist sub-agents
    • • Pre-configured MCPs
    • • Pattern recall on every run
    Second Brain 2.0

    The complete Claude Code system. 30+ agents, memory that compounds, pre-configured MCPs.

    DIY $197 · Kickstart $597 · DWY $2,497

    See Packages
    MemoryOS

    The memory layer that makes discipline compound. Patterns persist across every future session.

    Free · Standard $199/yr · Pro $349/yr

    Get MemoryOS