agentleFS
calvyntwhcalvyntwh/karu-custom-skills/skills/pareto-principle/SKILL.md

pareto-principle

Skill1 starsChanged 2 months agoMITView on GitHub

Strategic Prioritization using the 80/20 Rule. Optimization of Scope. Use when planning tasks, managing feature creep, or maximizing ROI.

1 other project uses a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

17 rules. About 0 are common, 17 are distinctive, and 0 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈17

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 0

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Planning Phase: When the task list is long (> 5 items).

    line 14
  2. Distinctive:

    Feature Creep: When a project feels "unfocused" or "heavy".

    line 15
  3. Distinctive:

    Single item: No prioritization needed.

    line 20
  4. Distinctive:

    Early-stage ambiguity: When Value cannot be estimated because requirements are unclear.

    line 22
  5. Distinctive:

    Do not ask "Can we do this?".

    line 28
  6. Distinctive:

    Before any scoring, verify the mindset shift:

    line 31
  7. Distinctive:

    If option selection → use decision-matrix instead.

    line 33
  8. Distinctive:

    Top 20% (High ROI): DO IT NOW. This is the "Vital Few".

    line 71
  9. Distinctive:

    Bottom 20% (Negative ROI): DELETE or archive.

    line 73
  10. Distinctive:

    After completing each Top 20% item:

    line 79
  11. Distinctive:

    Ask: "Has context shifted?

    line 81
  12. Distinctive:

    Re-cut frequency: After each Top 20% completion, or bi-weekly for long-running projects.

    line 88
  13. Distinctive:

    Before labeling an item DELETE, ask:

    line 92
  14. Distinctive:

    If yes to any → Move to DEFER instead.

    line 97
  15. Distinctive:

    Before labeling an item DEFER, ask:

    line 99
  16. Distinctive:

    Before the final cut:

    line 105
  17. Distinctive:

    Defer Watch: If an item has been "DEFER" for > 3 months, explicitly ask user: "Is this still DEFER or should it be DELETE?"

    line 134

≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

The Pareto Principle (The Scope Hammer)

"80% of consequences come from 20% of the causes."

When to Use

  • Planning Phase: When the task list is long (> 5 items).
  • Feature Creep: When a project feels "unfocused" or "heavy".
  • Refactoring: Identifying which module is causing 80% of the bugs.

When NOT to Use

  • Single item: No prioritization needed.
  • Short list (< 5 items): Simple ranking is faster and sufficient.
  • Early-stage ambiguity: When Value cannot be estimated because requirements are unclear.
  • Time-critical delivery: The overhead of scoring is not worth it for one-off decisions.
  • High-stakes irreversible decisions: 80/20 is a heuristic; use decision-matrix for decisions that are hard to reverse.

The Protocol: The Scope Hammer

Do not ask "Can we do this?". Ask "Should we do this?".

Step 0: Pre-Check (Should We?)

Before any scoring, verify the mindset shift:

  • Is this about scope optimization (what to prioritize) or option selection (which one to choose)?
  • If option selection → use decision-matrix instead.

Step 1: List & Rate

List every proposed feature or task. Ask the User (or use Proxy Metrics):

  • Value (V): User rating 1-10 or ranked.
  • Effort (E): Proxy: file count, complexity, dependencies.

[!IMPORTANT] The V/E formula is a tiebreaker, not the core insight. If you can identify the vital few without it, skip the formula.

Step 2: Add Confidence Modifiers

Rate confidence in each estimate (1-3) on two separate axes:

V_conf (Value confidence) — how sure are you that this item creates the value you estimate?

V_confAnchorSource
1Speculative — gut feel, no evidenceJuran Quality Handbook: no data, opinion only
2Partial data — analogous case, anecdotal user research, or one sourceQSM Almanac: single-source estimate
3Measured — benchmarks, A/B test data, repeated user researchQSM Almanac: multi-source validated

E_conf (Effort confidence) — how sure are you that the effort estimate is accurate?

E_confAnchorSource
1Speculative — no prior work, novel stackQSM Almanac: unknown unknowns dominant
2Partial data — analogous feature shipped, partial decompositionQSM Almanac: range estimate plausible
3Measured — same feature shipped before, effort trackedQSM Almanac: historical data anchors

Apply discount: Adjusted Score = (V/E) × (V_conf × E_conf) / 4

Note on anchors: V_conf and E_conf use the same 1-3 scale but different anchoring rubrics. V_conf asks "do I have evidence this creates value?"; E_conf asks "do I have evidence this takes the effort I think?" Confusing the two leads to false precision.

Step 3: The Cut (Iterative, Not One-Time)

Top 20% (High ROI): DO IT NOW. This is the "Vital Few". Middle 60%: DEFER. Do not touch until Top 20% is complete. Bottom 20% (Negative ROI): DELETE or archive.

[!WARNING] DEFER is not PERMANENT. Items in the middle 60% should be re-evaluated after each Top 20% item is completed. The backlog changes; so should the prioritization.

Step 4: Re-Cut Trigger

After completing each Top 20% item:

  1. Recalculate ROI for remaining items (dependencies may have changed)
  2. Ask: "Has context shifted? Any deferred items now have higher Value?"
  3. Re-cut if necessary

Timeboxing

  • Maximum list size before simplification: 20 items. Beyond this, use grouping.
  • Scoring time budget: 5 minutes per item. If V or E cannot be estimated in that time, default to V=5, E=5 (score=1) and mark as uncertain.
  • Re-cut frequency: After each Top 20% completion, or bi-weekly for long-running projects.

Defer/DELETE Safeguard

Before labeling an item DELETE, ask:

  • "Will deleting this increase effort on remaining items?" (dependency)
  • "Will this become more expensive to build later?" (technical debt)
  • "Is this a 'trivially deferred' item that will become critical?" (scope creep pattern)

If yes to any → Move to DEFER instead.

Before labeling an item DEFER, ask:

  • "Can this be completed in < 2 hours?" If yes, just do it now.
  • "Is this a dependency blocker for Top 20% items?" If yes, elevate it.

Validation Steps

Before the final cut:

  1. Confirm distribution: Does 80/20 actually hold in your domain? Validate with historical data if possible.
  2. Check non-negotiables: Does any item violate a hard constraint (security, compliance)? Elevate regardless of ROI.
  3. Saboteur check: inversion-thinking - How could this prioritization backfire?

Skill Integration

SituationUse Instead/Also
Choosing between optionsdecision-matrix
Need to verify value estimatesmap-vs-territory
Criteria might be wrongchestertons-fence
Risk of over-analysisoccams-razor
Preventing deferral from becoming permanentinversion-thinking
Understanding cascading effectssecond-order-thinking

Self-Improvement Protocol

Log only if prediction was wrong.

## [YYYY-MM-DD] {Brief Description}
**Action:** {what was prioritized / deferred / deleted}
**Outcome:** {did it work?}
**Lesson:** {one sentence}

Review Trigger: Monthly → check CORRECTIONS.md → promote validated patterns to LEARNINGS.md.

Defer Watch: If an item has been "DEFER" for > 3 months, explicitly ask user: "Is this still DEFER or should it be DELETE?"

Resources

  • Detailed Research Notes
  • Impact Matrix Reference

Evaluations

Eval 1: Option Selection vs Scope Optimization

Scenario: User asks "React vs Vue for our startup." Expected: Identifies as option selection, redirects to decision-matrix (not pareto). Pass criteria: Correctly distinguishes "what to prioritize" from "which to choose", applies right skill.

Eval 2: Low-Effort Quick Win Detection

Scenario: List has 10 items. One item is "Update README" with V=5, E=1. Expected: Identifies as Top 20%, flags as "do in < 2 hours" for immediate execution. Pass criteria: Catches quick win that would be lost in formal scoring, recommends doing now.

Eval 3: Defer Permanence Prevention

Scenario: Feature has been "DEFER" for 3+ months. Expected: Flags for explicit user decision (DELETE or keep deferred), prevents permanent deferral. Pass criteria: Asks user explicitly after 3 months, does NOT allow indefinite deferral.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.