Skip to content

feat(skill-creator): encode progressive disclosure model#180

Merged
notque merged 1 commit intomainfrom
feat/skill-creator-progressive-disclosure
Mar 27, 2026
Merged

feat(skill-creator): encode progressive disclosure model#180
notque merged 1 commit intomainfrom
feat/skill-creator-progressive-disclosure

Conversation

@notque
Copy link
Copy Markdown
Owner

@notque notque commented Mar 27, 2026

Summary

  • Removes old Operator Context / CAN/CANNOT template from skill-template.md — replaced with workflow-first inline constraints
  • Creates references/progressive-disclosure.md — the core teaching document for the orchestrator+references model
  • Updates SKILL.md with 500/700-line extraction gates and "Maximizing skill effectiveness" guidance

What progressive-disclosure.md teaches

  • The four-directory model (SKILL.md + references/ + scripts/ + agents/)
  • Token economics: 300 lines to consider vs 1100 to execute
  • Real examples table from 6 toolkit skills (comprehensive-review through voice-writer)
  • Deterministic script principle: if it can be a script, it should be
  • Bundled agents pattern for parallel specialist dispatch
  • 6-step checklist for applying the model

Test plan

  • 732 tests pass locally
  • No old Operator Context sections remain as templates to teach
  • Progressive disclosure gate (500/700 lines) added to SKILL.md workflow

…-first template

Add references/progressive-disclosure.md teaching the lean-SKILL.md / rich-references
model: economics of routing-time vs execution-time loading, 500/700-line size gates,
what to extract (checklists, agent prompts, templates, rubrics), deterministic script
principle, bundled agents pattern, and a real-numbers table from 6 toolkit skills.

Remove the old Operator Context section from skill-template.md (separate
Hardcoded/Default/Optional Behaviors blocks) and replace with workflow-first
Instructions template showing inline constraints with "because" reasoning and
gate annotations where they apply.

Strengthen progressive disclosure guidance in SKILL.md: replace 3-line mention
with a full subsection referencing the new file, explicit 500/700-line gates,
and a "Maximizing skill effectiveness" table encoding more-references = better,
deterministic scripts preferred, bundled agents useful.
@notque notque merged commit 211b321 into main Mar 27, 2026
4 checks passed
@notque notque deleted the feat/skill-creator-progressive-disclosure branch March 27, 2026 19:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant