harnesskit

harnesskit is a cursor rule for Cursor from Oriolshhh/runware-image-mcp. It costs 2,956 tokens per session, scanned A, original, MIT.

A repository work-harness configuration that defines how coding agents discover context, route work, and validate changes. It uses project files such as YAML gates and generated rule files.

In plain words
What is it for?
It is for setting up repository guidance, context-loading rules, agent roles, model routing, and project validation commands.
Why use it?
It helps keep agent work consistent across a codebase and prevents broad, unnecessary exploration. It also gives the project a place to record checks that changes must pass.

Cursor rule for Cursor

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 rules/oriolshhh/runware-image-mcp/harnesskit
Clone the repo
git clone --depth 1 https://github.com/Oriolshhh/runware-image-mcp

Made for: Cursor.

Wrote this? Show the measurements

A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.

agentmods badge for harnesskit

README.md
[![agentmods](https://agentmods.dev/badge/rules/oriolshhh/runware-image-mcp/harnesskit.svg)](https://agentmods.dev/rules/oriolshhh/runware-image-mcp/harnesskit)
Your own site
<a href="https://agentmods.dev/rules/oriolshhh/runware-image-mcp/harnesskit"><img src="https://agentmods.dev/badge/rules/oriolshhh/runware-image-mcp/harnesskit.svg" alt="Measured on agentmods" height="20"></a>
Per session 2,956 This file is loaded in full into every session.
When invoked 2,956 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 5.1 $0.02956 $0.02956
Opus 5 $0.01478 $0.01478
Sonnet 5 $0.00591 $0.00591
Haiku 4.5 $0.00296 $0.00296

Measured yesterday against content hash b449aec2ceca, method: parsed. Prices are Anthropic first-party input rates as of 2026-09-05, from the pricing page.

Security

Grade A, and why

harnesskit 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 yesterday.

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.

.cursor/rules/harnesskit.mdc · 225 lines

How it starts

The opening of the file, as written. The whole thing — 225 lines — stays where its author put it; the contents beside it link to each section on GitHub.

harnesskit-workspace — HarnessKit core rules

How to use this file

  • .agent/harness.yml is the source of truth. Do not edit generated files; edit the source and run harnesskit export.
  • Apply the context-discovery skill before broad repository exploration: read .agent/context/README.md and .agent/context/routing.md when present, assess revision/scope freshness, then load only task-relevant context and verify critical claims against source. Coordinators should pass one concise context capsule to subagents so they do not all reload the same pack.
  • Component frontmatter is cheap discovery metadata. Agents, commands, and loops also declare model tier/reasoning; agents list skills to preload. Apply model-routing, load only relevant components, and do not load everything by default.
  • Agents are logical roles, not guaranteed separate processes.
  • Packs are composable behavior layers applied in priority order.
  • Correctness and safety always override style packs.
  • Explicit user instructions override pack preferences.
  • Non-trivial changes should follow the spec-driven workflow.

Agents

  • accessibility-reviewer — Independently reviews semantics, keyboard and focus behavior, contrast, motion, and recovery.
  • agent-builder — Creates focused, single-responsibility agent definitions.
  • context-engineer — Builds an evidence-backed repository context pack for new agents.
  • council-chair — Synthesizes specialist reports and rebuttals into one recommendation.
  • council-router — Selects the smallest set of relevant council specialists and justifies each.
  • debugger — Reproduces failures, proves root causes, applies minimal fixes, and verifies regressions.
  • design-system-specialist — Finds and extends existing tokens, primitives, components, and frontend conventions.
  • devils-advocate — Challenges the leading council proposal with evidence-based objections.
  • frontend-design-lead — Converts product intent into an intentional visual direction and implementation brief.
  • frontend-implementer — Implements approved frontend direction with responsive states, accessibility, and focused code.
  • implementer — Implements approved specs with tests and passing gates.
  • interaction-design-specialist — Specifies complete control feedback, interaction states, transitions, and purposeful motion.
  • motion-design-reviewer — Reviews whether interface motion is purposeful, responsive, accessible, and performant.
  • operations-reviewer — Council reviewer for rollout, migration, rollback, performance, and maintenance.
  • product-discovery — Turns an ambiguous request into a normalized, evidence-backed decision brief.
  • product-reviewer — Council reviewer for user value, scope, adoption, and problem fit.
  • reviewer — Reviews diffs against the spec and repository conventions.
  • security-reviewer — Council reviewer for trust boundaries, validation, secrets, and abuse cases.
  • solution-architect — Council reviewer for boundaries, APIs, data flow, compatibility, and alternatives.
  • spec-engineer — Adaptively coordinates direct, specialist-assisted, or council-backed specs and waits for approval.
  • task-router — Classifies open-ended requests and routes them to the smallest capable workflow or specialist.
  • test-strategist — Council reviewer for testability, acceptance criteria, and required gates.
  • tester — Designs and strengthens tests around expected behavior.
  • ux-reviewer — Performs actionable UX and accessibility review.
  • visual-assets-director — Finds purposeful image opportunities and defines coherent generation-ready asset systems for existing pages.
  • visual-craft-specialist — Audits and specifies product-specific typography, color, spacing, composition, imagery, and control polish.
  • visual-qa-tester — Exercises the rendered UI across states and viewports and records browser evidence.

Read the full file on GitHub · 225 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. yesterday First seen · 225 lines · 2,956 tokens per session scan A b449aec2ceca

Subscribe to this mod's changes

harnesskit is a cursor rule published in the GitHub repository Oriolshhh/runware-image-mcp (0 stars, last pushed 1mo ago), licensed MIT. It adds 2,956 tokens to every session, about $0.0148 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 2026-09-03.