autonomous-agent-patterns

Design patterns for building autonomous coding agents. Covers tool integration, permission systems, browser automation, and human-in-the-loop workflows. Use when building AI agents, designing tool ...

Install with Tessl CLI

npx tessl i github:sickn33/antigravity-awesome-skills --skill autonomous-agent-patterns

What are skills?

Review — 66%

Does it follow best practices?

If you maintain this skill, you can automatically optimize it using the tessl CLI to improve its score:

npx tessl skill review --optimize ./path/to/skill

Learn more

Evaluation — 98%

Agent success when using this skill

Validation — 9 / 11 Passed

Validation for skill structure

SKILL.md

Review

Evals

Evaluation results

100%

17%

Autonomous Coding Agent: Core Loop and Tool Implementation

Agent loop and multi-model routing

Criteria

Without context

With context

Max iterations cap

100%

Tool choice auto

100%

Fast model for planning

100%

Smart model for analysis

100%

Code model for generation

100%

Edit tool search/replace

100%

Expected occurrences validation

50%

100%

Read tool line range

100%

ToolResult on failure

100%

History tracking

100%

Tool registry dict

100%

Termination on no tool calls

50%

100%

Without context: $0.3086 · 1m 29s · 11 turns · 60 in / 5,884 out tokens

With context: $0.6236 · 1m 59s · 24 turns · 349 in / 7,461 out tokens

100%

AI Agent Permission and Sandboxing Framework

Permission system and sandboxed execution

Criteria

Without context

With context

Four permission levels

87%

100%

Read ops = AUTO

100%

Write ops = ASK_ONCE

100%

Destructive ops = ASK_EACH

100%

Dangerous ops = NEVER

100%

Session approval cache

100%

Risk assessment HIGH flags

80%

100%

Allowed commands whitelist

100%

Realpath path validation

100%

HOME env isolation

100%

NEVER blocks execution

100%

Subprocess timeout

100%

Without context: $0.3950 · 1m 54s · 13 turns · 61 in / 7,667 out tokens

With context: $0.4711 · 1m 34s · 19 turns · 65 in / 5,403 out tokens

95%

20%

AI Agent Context Management and Dynamic Tool Factory

Context management and MCP tool creation

Criteria

Without context

With context

@file method

100%

@folder method

100%

@url method

100%

@problems method

100%

Folder max_files cap

50%

Checkpoint saves git_ref

100%

Checkpoint saves git_dirty

70%

100%

Checkpoint restore

100%

FastMCP framework

100%

MCP hot-reload

100%

Context format_for_prompt

100%

Checkpoint storage dir

100%

Without context: $0.5969 · 2m 32s · 19 turns · 68 in / 10,710 out tokens

With context: $0.6265 · 1m 50s · 24 turns · 71 in / 6,886 out tokens

Evaluated: 2 days ago
Agent: Claude Code

Table of Contents

Autonomous Coding Agent: Core Loop and Tool Implementation AI Agent Permission and Sandboxing Framework AI Agent Context Management and Dynamic Tool Factory

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.