Skip to content
cccskills
FreePublished on GitHub

agent-harness-construction

Design and optimize AI agent action spaces, tool definitions, and observation formatting for higher completion rates.

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data
See how to install

Files included (2)

  • SKILL.md2.0 KB
  • metadata.json853 B

SKILL.md (original)

Review the instructions the agent will receive before you install.

Agent Harness Construction

Use this skill when you are improving how an agent plans, calls tools, recovers from errors, and converges on completion.

Core Model

Agent output quality is constrained by:

  1. Action space quality
  2. Observation quality
  3. Recovery quality
  4. Context budget quality

Action Space Design

  1. Use stable, explicit tool names.
  2. Keep inputs schema-first and narrow.
  3. Return deterministic output shapes.
  4. Avoid catch-all tools unless isolation is impossible.

Granularity Rules

  • Use micro-tools for high-risk operations (deploy, migration, permissions).
  • Use medium tools for common edit/read/search loops.
  • Use macro-tools only when round-trip overhead is the dominant cost.

Observation Design

Every tool response should include:

  • status: success|warning|error
  • summary: one-line result
  • next_actions: actionable follow-ups
  • artifacts: file paths / IDs

Error Recovery Contract

For every error path, include:

  • root cause hint
  • safe retry instruction
  • explicit stop condition

Context Budgeting

  1. Keep system prompt minimal and invariant.
  2. Move large guidance into skills loaded on demand.
  3. Prefer references to files over inlining long documents.
  4. Compact at phase boundaries, not arbitrary token thresholds.

Architecture Pattern Guidance

  • ReAct: best for exploratory tasks with uncertain path.
  • Function-calling: best for structured deterministic flows.
  • Hybrid (recommended): ReAct planning + typed tool execution.

Benchmarking

Track:

  • completion rate
  • retries per task
  • pass@1 and pass@3
  • cost per successful task

Anti-Patterns

  • Too many tools with overlapping semantics.
  • Opaque tool output with no recovery hints.
  • Error-only output without next steps.
  • Context overloading with irrelevant references.

Reviews

No reviews yet. Tell others what you thought.

Comments

No login needed. Share how it worked for you, ask a question, or note where it did not run.

No comments yet. Be the first to write one.

Comments are published right away. Do not include personal information such as email addresses or phone numbers. Up to 2 URLs.

0 / 1,000 characters

This form is protected by reCAPTCHA, and the Google Privacy Policy and Terms of Service apply.

More skills from this repository

003-agents-installation

Free

Use when you need to install the embedded robot agents into either .cursor/agents or .claude/agents, selecting the destination interactively and copying the embedded agent definitions from project assets. This should trigger for requests such as Install embedded agents; Bootstrap .cursor/agents; Bootstrap .claude/agents; Copy robot agents. Part of cursor-rules-java project

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

200-agents-md

Free

Use when you need to generate an AGENTS.md file for a Java repository — covering project conventions, tech stack, file structure, commands, Git workflow, and contributor boundaries — through a modular, step-based interactive process that adapts to your specific project needs. This should trigger for requests such as Create AGENTS.md; Update AGENTS.md file; Add agent instructions. Part of cursor-rules-java project

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

2026-legal-research-agent

Free

Expert legal research agent for finding and scraping expungement data state by state. Knows authoritative sources, URL patterns, Firecrawl configuration, and 2026 legal landscape. Activate on "find expungement data", "scrape state laws", "legal research", "court URLs", "statute sources", "Clean Slate laws", "automatic expungement research". NOT for interpreting laws (use national-expungement-expert), building UI, or legal advice.

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

20260102-214532-trinity-auto-boot-validator

Free

Generated by Gremlin Forge Meta-Agent 🍆👾⚡

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

80-01-autonomous-work

Free

Work autonomously without waiting for hand-holding — complete tasks end-to-end without asking for permission at every step.

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

abmahnung

Free

Entwurf einer BAG-konformen Abmahnung im Arbeitsrecht – konkrete Schilderung der Pflichtverletzung, klare Verhaltenserwartung, ausdrückliche Sanktionsandrohung. Use when AG eine verhaltensbedingte Sanktion vor einer Kündigung dokumentieren muss oder AN die Wirksamkeit einer empfangenen Abmahnung prüft.

The English summary is being prepared. Showing the original description.

majiayu000/claude-skill-registry-data27Updated Oct 21, 2026

See all skills by majiayu000

Report a problem with this skill