Public: Claude Code as an agent harness
researchedcoding-assistant
framework
Agent Framework
harness
Harness / IDE
memory
Memory
CC
claude-code-memory
eval
Eval
UC
user-configurable
observability
Observability / Tracing
UC
user-configurable
sandbox
Sandbox / Code Execution
UC
user-configurable
deployment
Deployment Target
LO
local
mcp_servers
MCP Servers
UC
user-configurable
Architectural notes
Claude Code itself is a documented agent harness with a public primitive set: skills (SKILL.md), hooks, subagents, plugins, marketplaces, MCP support, and a project memory system. Notable architectural choices (from official docs): - Skills standard (SKILL.md) shared with OpenAI Codex CLI. - Plugin marketplace model with version pinning. - Hooks for PreToolUse / PostToolUse / Stop events. - Auto-memory system for persistent user/project context. Useful as the canonical reference for "harness-led" agent stacks where the harness itself supplies most of the platform.
Sources
Embed this stack
Show off your stack on your engineering blog or company page.
<iframe src="http://testserver/embed/stacks/claude-code-agentic" width="100%" height="380" frameborder="0" style="border: 1px solid #334155; border-radius: 8px;"></iframe>