claude-code-harness AGENTS.md

Project instructions for maintaining the claude-code-harness repository, which contains two Claude Code plugins and their tests. They describe the repository layout, commands, validation rules, and file boundaries.

In plain words
What is it for?
Use them when editing the harness or devlog plugins, checking manifests, running tests, and preparing releases.
Why use it?
They help contributors change the correct plugin files and avoid confusing the repository's own development setup with files that ship in the plugins.

Instructions file for CodexOpenCode

Install

Getting it into your agent

One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.

agentmods
npx agentmods add instructions/nikitacodesave/claude-code-harness/agents-md
Clone the repo
git clone --depth 1 https://github.com/nikitaCodeSave/claude-code-harness

Made for: Codex, OpenCode.

Per session 1,300 This file is loaded in full into every session.
When invoked 1,300 The same file — it is already loaded in full.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

What it costs to keep this loaded

Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.

ModelPer sessionOnce invoked
Fable 5 $0.01300 $0.01300
Opus 5 $0.00650 $0.00650
Sonnet 5 $0.00260 $0.00260
Haiku 4.5 $0.00130 $0.00130

Measured 2d ago against content hash f2c4fee35fe1, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

claude-code-harness AGENTS.md scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2d ago.

A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.

Nothing flagged

None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.

AGENTS.md · 90 lines

How it starts

The opening of the file, as written. The whole thing — 90 lines — stays where its author put it; the contents beside it link to each section on GitHub.

claude-code-harness — operational notes for coding agents

This repository is the source of truth for two Claude Code plugins. Read this before changing anything; it carries what the file tree cannot tell you. Claude Code additionally loads .claude/CLAUDE.md (maintainer-facing, Russian); agents from other vendors get this file.

Layout — the part that is easy to break

  • .claude-plugin/marketplace.json (repo root) is the marketplace catalog. Each plugin's own manifest is plugins/<name>/.claude-plugin/plugin.json.
  • There are two plugins: plugins/harness/ (the kit — a skill plus on-demand references) and plugins/devlog/ (the continuity companion the kit points at).
  • .claude/ at the repo root is this repository's own dev harness. It is not part of either plugin and never ships.
  • ~/.claude/skills/claude-code-harness on the maintainer's machine is a symlink to plugins/harness/, so edits here are live immediately. Edit here, never in a copy.
  • evals/ is the kit's own regression suite — mechanical checks plus fixture projects and a criteria file written before the runs. Dev harness, never shipped.

Commands

# release: writes the version into BOTH plugin.json files in lockstep, checks shipped-by stamps,
# stages the shippable surface, prints the commit/tag/push finish. It never commits for you.
./plugins/harness/scripts/release.sh <version>

# manifest validation
claude plugin validate .

# devlog index (required after adding an entry under .claude/devlog/entries/)
./plugins/devlog/bin/devlog-reindex

# regression suite — seconds, free; run it before every release
./evals/run.sh mech

Done = ./evals/run.sh mech exits 0, release.sh completes without a stamp mismatch, plugin validate prints ✔ Validation passed, and the reindexer reports no ERRORS. The suite was red for four consecutive releases because this line did not name it.

Boundaries

  • Always: bump the version in plugin.json for any change consumers should receive — it is the update cache key, so pushing content without bumping it ships nothing. Use release.sh; two manual releases in a row shipped desynced manifests, which is why the ritual is a script.
  • Ask first: changing what the plugin ships (adding an agent, a command, a hook). The kit's own position is that it ships a skill and references only; adding machinery needs a reason that survives the question "does a built-in already do this?"
  • Never: edit .claude/devlog/entries/ (existing entries) or .claude/audits/ — frozen provenance. Never hand-edit index.json / tldr.md; they are generated. Never put a version field in the marketplace entry — plugin.json is authoritative and the release script fails the build if one appears.

Read the full file on GitHub · 90 lines

Changes

What this file has done since we first saw it

Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.

  1. 2d ago First seen · 90 lines · 1,300 tokens per session scan A f2c4fee35fe1

Subscribe to this mod's changes

claude-code-harness AGENTS.md is an instructions file published in the GitHub repository nikitaCodeSave/claude-code-harness (5 stars, last pushed 3d ago), licensed MIT. It adds 1,300 tokens to every session, about $0.0065 per session on Opus 5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.

Related

Other instructions, from other repositories

vscode buildNext.instructions.md

Working notes and architecture documentation for the new esbuild-based build system in build/next. Use when making changes to the new build pipeline (transpile/bundle commands, NLS plugin, source-map handling, resource copying, or self-hosting watch tasks).

microsoft/vscode · 6,785 tokens

spec-kit AGENTS.md

AGENTS.md instructions for github/spec-kit, covering agents.md, about spec kit and specify, quickstart — add a new integration in 5 steps, integration architecture and integrationmanifest — file tracking.

github/spec-kit · 7,104 tokens

codex AGENTS.md

AGENTS.md instructions for openai/codex, covering rust/codex-rs, the codex-core crate, code review rules, crate api surface and model visible context.

openai/codex · 5,182 tokens

langchain AGENTS.md

AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.

langchain-ai/langchain · 4,345 tokens

vscode oss-third-party-notices.instructions.md

Instructions for microsoft/vscode, covering vs code oss third-party-notices pipeline, architecture, pipeline flow in ci, applying the notice (cutover) and fallback chain (never fail the build).

microsoft/vscode · 5,001 tokens

next.js AGENTS.md

Instructions for vercel/next.js, covering next.js development guide, codebase structure, monorepo overview, core package: packages/next and other important packages.

vercel/next.js · 7,296 tokens