sblattj

Agentic Coding Harness — Development skill for Claude Code

Development community

Run, watch, and meter Claude Code, Codex CLI, Gemini CLI, OpenCode, and Kiro from one CLI.

How to install Agentic Coding Harness

This entry records only its repository, not the path inside it, so there is no exact command to give. Open sblattj/agentic-coding-harness and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agentic Coding Harness does

Run, watch, and meter Claude Code, Codex CLI, Gemini CLI, OpenCode, and Kiro from one CLI. Normalized events, cache-aware token usage, cost accounting verified against each agent's own records, budget caps, OpenTelemetry gen_ai spans, Langfuse traces, and a live web dashboard + MCP server.

Alternatives in Development

10 Colonial Records Search — Colonial Records Search 1.1k ★
Projectmem — Open-source coding agent memory 766 ★
Project Memory: AgentSys — Cross-tool compatible: Works across all 5 platforms - Claude Code, OpenCode, Codex CLI, Cursor, and Kiro 622 ★

README

agentic-coding-harness — cost tracking and observability for Claude Code, Codex CLI, Gemini CLI, OpenCode, and Kiro

"node =18.19" src="https://img.shields.io/badge/node-%3E%3D18.19-brightgreen">

**Agents hide the burn. This is the receipt — tokens, credits, and dollars, verified against each CLI's own records.**

`agentic-coding-harness` (`ach`) runs, watches, and meters AI coding agents from one CLI. Launch headless runs on **Claude Code, Codex CLI, Gemini CLI, OpenCode, or Kiro** and get token usage — input, output, cache read, cache write, reasoning — plus per-model cost, Kiro metering credits, and duration for every run. Costs are not estimated guesses: the cache-aware accounting is verified against each CLI's own ground-truth records (Claude session JSONL, Kiro session files, provider-reported `costUSD`). Every run persists as a replayable artifact — an ATIF trajectory, OpenTelemetry `gen_ai` spans, a Langfuse trace, or a single-file HTML report — and a live terminal dashboard, web dashboard, and MCP server put usage, cost, and credits for all five agents on one screen.

![Live web dashboard grid showing token usage and cost across concurrent Claude Code and Kiro agent runs, with interactive terminal panes](docs/assets/dashboard.gif)

npm install -g agentic-coding-harness    # or: pip install agentic-coding-harness
ach run --agent claude "add a --dry-run flag to scripts/lint.sh"
ach web                                  # live dashboard on :8399