agentleFS
komlukkomluk/scaffolding/skills/context-engineering/SKILL.md

context-engineering

Skill16 starsChanged 4 days agoMITView on GitHub

Optimize Claude Code context-window usage for accuracy and cost. TRIGGER when: hitting context limits, structuring prompts for an agent, or trimming what gets injected into a task. SKIP: persisting knowledge across sessions (use agent-memory); semantic recall tuning (use semantic-memory-mcp).

131 other projects use a version of this.

Adopted by

Most recently taken up by

Used with

Projects with it also use using-agent-skills, idea-refine and 10 more.

Other files people use alongside this one.

Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
  • LLl0k0laDSkill

    using-agent-skills

    Use when starting a session or when you need to discover which skill applies to the current task. This is the meta-skill that governs how all other skills are discovered and invoked.

    Also in 36 projects that use this

  • LLl0k0laDSkill

    idea-refine

    Refine ideas through structured divergent and convergent thinking. Use "idea-refine" or "ideate" to trigger.

    Also in 44 projects that use this

  • HK-hubSkill

    deprecation-and-migration

    Manages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when deciding whether to maintain or sunset existing code.

    Also in 40 projects that use this

  • MrCipherSmithSkill

    constraint-driven-development

    Use when a project needs a persistent, measurable quality bar — starting a project without documented standards, 'set up constraints', or when coverage/perf/a11y thresholds get re-debated per-PR. Produces CONSTRAINTS.md with enforceable numbers, ratchets, and a guard against silent weakening. NOT for throwaway prototypes or a one-off CI setup.

    Also in 14 projects that use this

  • HK-hubSkill

    source-driven-development

    Grounds every implementation decision in official documentation. Use when you want authoritative, source-cited code free from outdated patterns. Use when building with any framework or library where correctness matters.

    Also in 44 projects that use this

  • +6 more from LLl0k0laD
  • +1 more from HK-hub

The file

We split this skill into the rules it gives an agent

6 rules. 0 are common, 6 are distinctive, and 0 are very specific to this project.

  • 0

    Common

    Most of the 117 edited versions we read keep these.

  • 6

    Distinctive

    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:

    Every 50 messages in conversation

    line 14
  2. Distinctive:

    When context feels stale

    line 16
  3. Distinctive:

    Keep only relevant information

    line 23
  4. Distinctive:

    Remove completed task details

    line 24
  5. Distinctive:

    Use tables instead of prose

    line 139
  6. Distinctive:

    Remove examples from main docs

    line 141

“Common” is measured: kept by at least 6 in 10 of the 117 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

Context Engineering Skill

Optimizes Claude Code context for better performance and accuracy.

Tracked files in repo: !git ls-files 2>/dev/null | wc -l || true

Auto-Invoke Triggers

  • Every 50 messages in conversation
  • After major refactoring
  • When context feels stale
  • Task switching
  • User requests /context command

Context Principles

Active Curation

  • Keep only relevant information
  • Remove completed task details
  • Archive outdated patterns
  • Focus on current work

Signal-to-Noise Ratio

High SignalLow Signal
Current APIsOld experiments
Active patternsCompleted tasks
Recent decisionsHistorical context
Project structureImplementation details

Token Budget

FileLimitPurpose
CLAUDE.md< 2000 tokensQuick reference
Agent docs< 500 per agentRole definition
Skills< 300 per skillMethodology

Context Analysis Process

Step 1: Measure Current State

## Context Analysis

### Token Counts
- CLAUDE.md: XXX tokens
- Agent docs: XXX tokens total
- Skills: XXX tokens total
- Total: XXXX tokens

### Staleness Check
- [ ] Last updated: [date]
- [ ] References current project state
- [ ] No deprecated APIs/patterns

Step 2: Identify Issues

### Issues Found

#### Stale Content
- Section X: Outdated [reason]
- Section Y: No longer relevant

#### Redundant Content
- Duplicated in files A and B
- Can consolidate X and Y

#### Missing Content
- Current API not documented
- New pattern not captured

Step 3: Optimization Actions

### Optimization Plan

1. **Archive**: Move section X to knowledge-core.md
2. **Remove**: Delete duplicate Y
3. **Update**: Refresh API documentation
4. **Add**: Document new pattern Z

CLAUDE.md Structure

Optimal structure for quick reference:

# Project Name

[One-line description]

## Quick Reference
| Action | Command |
|--------|---------|
| Run | command |
| Test | command |
| Build | command |

## Agent System
[Link to detailed docs]

## Tech Stack
- Frontend: [stack]
- Backend: [stack]

## Project Structure
[Key directories only]

## Key Rules
1. [Critical rule]
2. [Critical rule]

Always-loaded vs lazy path-scoped rules

CLAUDE.md content splits into two tiers by when it must be available:

TierLives inLoadsExamples
Always-loadedproject-root CLAUDE.mdevery message, before any file is touchedrouting Protocol + Decision Tree + Agents table
Lazy / path-scopednested CLAUDE.md in a subdironly while editing under that directory treeper-area code conventions, stack notes, types location

Hard rule: the routing sections NEVER move into a nested CLAUDE.md — routing must fire on the first message, but a nested file only loads on Edit/Write under its tree (too late to route). Only path-specific code conventions are candidates to relocate, and only via the opt-in /init-rules scaffolder. This trims always-loaded tokens (modest for the plugin CLAUDE.md, larger for heavy project repos) without weakening routing. /doctor asserts the routing section is still present in the root CLAUDE.md.

Optimization Strategies

Token Reduction

  • Use tables instead of prose
  • Link to details instead of inline
  • Remove examples from main docs
  • Keep explanations brief

Information Hierarchy

  1. Immediate: In CLAUDE.md
  2. Reference: In linked docs
  3. Archive: In knowledge-core.md
  4. Delete: Truly obsolete

Import Chain Management

  • Max 5 levels of imports
  • No circular references
  • Clear dependency direction

Context Health Metrics

MetricHealthyWarningCritical
CLAUDE.md tokens< 15001500-2500> 2500
Stale sections01-23+
Redundancy0%< 10%> 10%
Missing critical012+

Maintenance Schedule

TriggerAction
Every 50 messagesQuick analysis
After major changeUpdate relevant sections
WeeklyFull review
Task switchClear task-specific context

Output: Context Report

## Context Report

**Date**: [date]
**Health**: Healthy | Warning | Critical

### Metrics
- Total tokens: XXXX
- Stale sections: X
- Redundancy: X%

### Findings
1. [Finding]
2. [Finding]

### Recommendations
1. [Action to take]
2. [Action to take]

### Potential Savings
- Current: XXXX tokens
- After optimization: XXXX tokens
- Savings: XX%

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.