agentleFS
kaka-rutokaka-ruto/skills/continuous-improvement/SKILL.md

continuous-improvement

Skill4 starsChanged 7 months agoMITView on GitHub

Use while building or maintaining a project to continuously improve the development system itself by fixing recurring process failures, tightening docs/skills, and proposing larger improvements when uncertain

5 other projects use a version of this.

Adopted by

Most recently taken up by

  • Other versions5 projects use a version of this.

    Word for word · 0

    None found.

    Edited · 0

    None found.

    Wrote their own version · 5

    1. l-gevity/l-gevity-skills★ 88same name
    2. muyen/vibe-to-prod★ 34same name
    3. naimkatiman/continuous-improvement★ 9same name
    4. BurakD/ai-engineering-harness★ 1same name
    5. Abdox-menara/guard-skills★ 0same name

    How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.

  • In this repository3 other files it gives its agents.

The file

We split this skill into the rules it gives an agent

11 rules. 0 are common, 8 are distinctive, and 3 are very specific to this project.

  • 0

    Common

    Most of the 4 edited versions we read keep these.

  • 8

    Distinctive

    This author's own choices. Worth a look.

  • 3

    Very specific

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

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    Always alongside active project work (features, fixes, refactors, docs)

    line 16
  2. Distinctive:

    Apply: update the smallest existing artifact that prevents recurrence.

    line 37
  3. Distinctive:

    Verify: ensure the change is consistent and doesn’t add busywork.

    line 38
  4. Distinctive:

    Treat as a pattern if any is true:

    line 46
  5. Very specific:

    Add or tighten a PLAN.md slice rule like “verification: required”

    line 63
  6. Distinctive:

    Add a missing “don’t print secrets” line to an existing runbook/skill

    line 65
  7. Distinctive:

    Add a “before you say done: run X” reminder to an existing checklist

    line 66
  8. Very specific:

    Do not “improve the system” by adding operating rules to PLAN.md.

    line 93
  9. Distinctive:

    Put durable process rules in AGENTS.md.

    line 94
  10. Distinctive:

    If you made any system improvement:

    line 105
  11. Very specific:

    If you proposed something, leave the proposal in the conversation (and optionally a note in PLAN.md)

    line 108

“Common” is measured: kept by at least 6 in 10 of the 4 edited versions of this file we could read. “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

Continuous Improvement

Overview

This skill makes the development system better over time. While doing normal project work, continuously look for repeated friction (missed checks, unclear steps, inconsistent standards, avoidable mistakes) and convert it into better defaults via small, safe process improvements.

Core principle: fix the system, not just the symptom.

When to Use

  • Always alongside active project work (features, fixes, refactors, docs)
  • Whenever you notice repeated misses like “forgot to run full tests”, unclear handoffs, or recurring production/debugging confusion

What This Skill Changes (And What It Doesn’t)

This skill can improve:

  • Existing docs in the repo (VISION.md, MISSION.md, PLAN.md, README/runbooks)
  • Existing global skills under ~/.agents/skills/
  • Existing project-level agent guidance files (like AGENTS.md), if present

This skill does not:

  • Silently introduce large process changes
  • Make destructive operational changes (especially in production)

The Improvement Loop (Always Running)

During work, continuously do:

  1. Notice: capture the friction in one sentence.
  2. Classify: is it a one-off or a pattern?
  3. Choose: fix now (small + certain) or propose (bigger/uncertain).
  4. Apply: update the smallest existing artifact that prevents recurrence.
  5. Verify: ensure the change is consistent and doesn’t add busywork.

Special Pattern: Agents Stopping With Work Remaining

If an agent repeatedly stops while PLAN.md Now still has unchecked items (and there is no hard blocker), treat it as a process failure and fix the system by tightening the relevant guidance (prefer AGENTS.md, then the continuous-development skill) so the default behavior is to pick the top-most item and continue.

Classification Rules

Treat as a pattern if any is true:

  • It happened more than once in the same session
  • It’s a common failure mode across agents/models (verification skipped, scope creep, missing handoff)
  • It’s high-impact even once (secrets exposure risk, production risk, data loss risk)

Otherwise treat as a one-off and leave a note in PLAN.md if needed.

Fix vs Propose (Decision Gate)

You may implement immediately (no ask) only when ALL are true

  • The change is small (few lines, localized)
  • It updates existing docs/skills/checklists (no new files)
  • It is clearly beneficial and low-risk (reduces ambiguity, adds a reminder, tightens a cap)
  • It does not meaningfully change product scope or engineering policy

Examples of safe immediate improvements:

  • Add or tighten a PLAN.md slice rule like “verification: required”
  • Clarify pruning caps in PLAN.md/MISSION.md
  • Add a missing “don’t print secrets” line to an existing runbook/skill
  • Add a “before you say done: run X” reminder to an existing checklist

You must propose first (and wait for approval) when ANY are true

  • Creating new docs/skills/templates
  • Editing CI configs (GitHub Actions, Buildkite, etc.)
  • Adding new checks that could slow development meaningfully
  • Changing team policy (branching strategy, release process, deployment rules)
  • Any change where you are not sure it will improve the system

When proposing, present:

  • The observed pattern (1–2 sentences)
  • The smallest change that prevents recurrence
  • The tradeoff (time/cost/noise)

Where to Put Improvements (Order of Preference)

  1. AGENTS.md (durable, agent-facing defaults; best for process rules)
  2. MISSION.md checklist wording (prevents scope drift)
  3. VISION.md clarifications (prevents direction drift)
  4. Existing repo docs/runbooks (README, troubleshooting guides)
  5. Existing global skills (when the fix is broadly reusable)
  6. PLAN.md task queue (transient; task items only)
  7. CI changes (only when approved and justified)

Special rule: keep PLAN.md free of operating rules

Do not “improve the system” by adding operating rules to PLAN.md. PLAN.md is for slices and status only. Put durable process rules in AGENTS.md.

Anti-Patterns to Avoid

  • Turning docs into verbose specs
  • Adding rules that are hard to follow or easy to ignore
  • Adding new checklists for everything instead of tightening the few that matter
  • “Process theater”: changes that look good but don’t prevent the actual failure

Minimal Output Standard (End of Session)

If you made any system improvement:

  • Mention what changed and why (1–2 sentences)
  • Point to the artifact changed (file path)
  • If you proposed something, leave the proposal in the conversation (and optionally a note in PLAN.md)

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.