skills.rest
Blake-John
Blake-JohnCOMMUNITY
Explore all skills by this author →
GitHub

agent-harness-construction

Design action spaces, observation formats, and error recovery contracts for AI agents.

Updated Mar 17, 2026
Search Tags:#planning#architecture#benchmarking#agent#error-recovery#tool-definition#observation-formatting
One-click install
npx skills add https://github.com/Blake-John/agent-config --skill agent-harness-construction-blake-john
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: agent-harness-construction
Source: https://github.com/Blake-John/agent-config/tree/main/.agents/skill_spec/agent-harness-construction
Command: npx skills add https://github.com/Blake-John/agent-config --skill agent-harness-construction-blake-john

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Agent Harness Construction helps designers improve how an AI agent plans, calls tools, recovers from errors, and converges on completion, increasing reliability and success rates.

Core Features & Use Cases

  • Action Space Design: Use stable, explicit tool names; inputs schema-first and narrow; return deterministic output shapes.
  • Granularity Rules: Use micro-tools for high-risk operations; use medium tools for common edit/read/search loops; use macro-tools only when round-trip overhead is the dominant cost.
  • Observation Design: Each 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: Keep system prompt minimal and invariant; move large guidance into skills loaded on demand; prefer references to files over inlining long documents; compact at phase boundaries.
  • 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.

Quick Start

Propose a redesigned action space, explicit tool definitions, and observation formatting to improve an agent’s completion rate on a given task.

Frequently Asked Questions about agent-harness-construction

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I improve AI agent planning and task completion rates?▼

Improve AI agent planning by applying schema-first action space design, explicit observation formatting, and error recovery contracts to enable reliable convergence and higher completion rates. Redesign tool definitions to include status, summary, next_actions, and artifacts for actionable feedback.

What is the best observation format for autonomous agent tool responses?▼

The best observation format for autonomous agent tool responses includes a status field, a one-line summary, actionable next_actions, and artifacts with file paths or IDs. This structured formatting provides explicit recovery hints and guides subsequent planning steps.

How do I design an error recovery contract for an autonomous agent?▼

Design an error recovery contract by defining a root cause hint, safe retry instruction, and explicit stop condition for every error path. This ensures the autonomous agent can handle failures gracefully without infinite loops or context overloading.

When should I use ReAct versus function-calling architecture for an AI agent?▼

Use ReAct architecture for exploratory tasks with an uncertain path, and function-calling for structured deterministic flows. A hybrid architecture is recommended, combining ReAct planning with typed tool execution to balance flexibility and reliability across agent tasks.

What metrics should I track when benchmarking autonomous agent reliability?▼

Track completion rate, retries per task, pass@1, pass@3, and cost per successful task when benchmarking autonomous agent reliability. These metrics quantify planning efficiency, error recovery effectiveness, and overall convergence speed across development and production environments.

Why does my AI agent fail to converge on task completion with too many tools?▼

An AI agent fails to converge on task completion when too many tools have overlapping semantics, creating opaque output with no recovery hints. Eliminate this anti-pattern by using explicit, narrow tools and context budgeting to compact references at phase boundaries.

Related Skills

Explore More Software Engineering Skills→
ValenBorrasValenBorras
1

agent-harness-construction

Design AI agent action spaces with explicit tool schemas and observation formats.

Community
Advanced
riftzen-bitriftzen-bit

agent-harness-construction

Design AI agent action spaces, tool definitions, and structured observation formats.

Community
Advanced
Ikram-AlamIkram-Alam

context-fundamentals

Organize prompts, tools, documents, and history into structured context components.

Community
Basic
Sahib-Sawhney-WHSahib-Sawhney-WH
4

agent-builder

Validate and configure DAPR Agents framework agent definitions and tools.

Community
Advanced