openclaw-conventions

openclaw-conventions is a cursor rule for coding agents from mergisi/openclaw-rules. It costs 468 tokens per session, scanned A, original, MIT.

A set of coding and project conventions for OpenClaw, covering file names, JavaScript scripts, agents, Git, secrets, and Telegram.

In plain words
What is it for?
Choosing names and directories, writing CommonJS scripts, handling HTTP requests and errors, organizing agents, protecting credentials, and formatting commits.
Why use it?
It prevents inconsistent project structure and common mistakes such as committing passwords, using the wrong module style, or naming files unpredictably.

Cursor rule

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/mergisi/openclaw-rules/openclaw-conventions
Clone the repo
git clone --depth 1 https://github.com/mergisi/openclaw-rules

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 openclaw-conventions

README.md
[![agentmods](https://agentmods.dev/badge/rules/mergisi/openclaw-rules/openclaw-conventions.svg)](https://agentmods.dev/rules/mergisi/openclaw-rules/openclaw-conventions)
Your own site
<a href="https://agentmods.dev/rules/mergisi/openclaw-rules/openclaw-conventions"><img src="https://agentmods.dev/badge/rules/mergisi/openclaw-rules/openclaw-conventions.svg" alt="Measured on agentmods" height="20"></a>
Per session 468 This file is loaded in full into every session.
When invoked 468 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 $0.00468 $0.00468
Opus 5 $0.00234 $0.00234
Sonnet 5 $0.00094 $0.00094
Haiku 4.5 $0.00047 $0.00047

Measured 3d ago against content hash 2bbdea8099ce, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

openclaw-conventions 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 3d 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.

global-rules/openclaw-conventions.mdc · 49 lines

What it actually says

OpenClaw Coding Conventions

General conventions for any OpenClaw project.

File Naming

  • Agent configs: config.json (not agent-config.json)
  • Identity: SOUL.md (always uppercase)
  • Memory: MEMORY.md (always uppercase)
  • Tool scripts: kebab-case.cjs (e.g., ga4-report.cjs, reddit-scan.cjs)
  • Secrets: secrets/ directory, never committed to git

JavaScript Style

  • CommonJS (.cjs), not ESM
  • No npm dependencies for tool scripts
  • Use https.request() not https.get() (more reliable)
  • Always call req.end() after https.request()
  • Buffer size: 512KB minimum for API responses
  • Rate limiting: Add sleep() between sequential API calls
  • Error handling: main().catch(e => { console.error("Fatal:", e.message); process.exit(1); })

Agent Naming

  • Agent names: Short, memorable, one word (Orion, Echo, Radar, Scout, Loki)
  • Session keys: agent:{name}:{scope} (lowercase)
  • Directories: agents/{name}/ (lowercase)

Git

  • Never commit secrets/, .env, API keys, or credentials
  • .gitignore should include: secrets/, .env, node_modules/, *.log
  • Commit messages: imperative mood ("Add report script" not "Added report script")

Telegram

  • Use plain text, never Markdown parse mode
  • Split messages over 4000 characters
  • Add 500ms delay between sequential messages
  • For copy-paste UX, send each item as a separate message

Common Pitfalls

  • https.get() can silently fail in script context — use https.request() + req.end()
  • Reddit JSON responses can be 150-200KB — set buffer to 512KB
  • Telegram Markdown parsing breaks on []_* in dynamic content — use plain text
  • Cron jobs need gateway restart to take effect
  • Path differences between Mac and Raspi — use config or env vars, not hardcoded paths
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. 3d ago First seen · 49 lines · 468 tokens per session scan A 2bbdea8099ce

Subscribe to this mod's changes

openclaw-conventions is a cursor rule published in the GitHub repository mergisi/openclaw-rules (2 stars, last pushed 6mo ago), licensed MIT. It adds 468 tokens to every session, about $0.0023 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.