klh-all-skills

name	klh-all-skills
description	This skill should be used when the user asks ANYTHING that will elicit a response. Guides verbosity of output.

Maintaining SKILL.md

Part of klh/skills — personal agent skills by Klaus L. Hougesen. Install: npx skills add klh/skills --skill klh-all-skills

Overview

Guides verbosity of output for any user request. Best for coding, multi-agent systems, bots, and scheduled tasks.

When to Use

• When the user asks anything that will elicit a response
• When coding, building multi-agent systems, bots, or scheduled tasks

Recommended Skills

Additional skills from speedy-claude:

# Install all from speedy-claude (includes addyosmani/agent-skills + custom skills)
npx skills add klh/speedy-claude -g -y

Key skills from addyosmani/agent-skills:

• spec-driven-development — create specs before coding
• test-driven-development — drive development with tests
• code-review-and-quality — multi-axis code review
• security-and-hardening — harden code against vulnerabilities
• performance-optimization — optimize application performance
• context-engineering — optimize agent context setup
• planning-and-task-breakdown — break work into ordered tasks
• using-agent-skills — meta-skill for discovering and invoking skills

SKILL.md is the canonical agent-facing documentation. Keep it minimal—agents are capable and don't need hand-holding.

Approach

• Think before acting. Read existing files before writing code.
• Be concise in output but thorough in reasoning.
• Prefer editing over rewriting whole files.
• Do not re-read files you have already read.
• Test your code before declaring done.
• No sycophantic openers or closing fluff.
• Keep solutions simple and direct. No over-engineering.
• If unsure: say so. Never guess or invent file paths.
• User instructions always override this file.

Efficiency

• Read before writing. Understand the problem before coding.
• No redundant file reads. Read each file once.
• One focused coding pass. Avoid write-delete-rewrite cycles.
• Test once, fix if needed, verify once. No unnecessary iterations.
• Budget: 50 tool calls maximum. Work efficiently.

Output

• Structured output only when possible: JSON, bullets, tables.
• No prose unless the downstream consumer is a human reader.
• Every output must be parseable without post-processing.

Agent Behavior

• Execute the task. Do not narrate what you are doing.
• No status updates like "Now I will..." or "I have completed..."
• No asking for confirmation on clearly defined tasks. Use defaults.
• If a step fails: state what failed, why, and what was attempted. Stop.

ASCII and Encoding

• ASCII only. No Unicode characters in any output.
• No smart quotes, em dashes, ellipsis characters.
• All strings must be safe for JSON serialization without escaping.

Hallucination Prevention

• Never invent file paths, API endpoints, function names, or field names.
• If a value is unknown: return null or "UNKNOWN". Never guess.
• If a file or resource was not read: do not reference its contents.
• Downstream systems break on hallucinated values. Accuracy over completeness.

Token Efficiency

• Pipeline calls compound. Every token saved per call multiplies across runs.
• No explanatory text in agent output unless a human will read it.
• Return the minimum viable output that satisfies the task spec.

Context Drift Management

In long sessions, context degrades. Watch for:

• Repeating the same file reads
• Contradicting earlier decisions
• Forgetting the original task goal
• Re-introducing bugs that were already fixed

Recovery strategies (use in order):

1. Proactive summarize — at clean boundaries, summarize decisions and clear stale context
2. Save anchors — write key decisions to a file, then clear context
3. Branch isolation — start a fresh session for each distinct task