Rule Creation

Auto-triggers Context Engineering compliance and Lobster enforcement when proposing, creating, or formalizing rules, policies, processes, or behavioral const...

Security Scan

The skill's code and instructions match its stated purpose (auto-wiring governance rules), but it is designed to write governance files immediately without asking and runs local commands—behavior that could be surprising or undesired if invoked autonomously.

Purpose & Capability

Name/description align with included scripts. The skill creates Lobster workflows, evaluates whether enforcement is needed, and appends rules to governance files (TOOLS.md, SOUL.md, etc.). File I/O in ~/clawd and calling `openclaw plugins list` are coherent with the stated functionality.

Instruction Scope

SKILL.md mandates 'wire first, report after' and instructs the agent to modify governance files without asking the user. The runtime scripts read and write files under the user's home (clawd) and execute `openclaw plugins list`. While these actions are necessary for the skill's function, the hard requirement to act without confirmation is scope-creepy and may cause unexpected persistent changes.

Install Mechanism

This is an instruction-and-script skill with no installer that pulls code from the package itself or a git clone. No network downloads or remote installers are performed by the code. Low install risk.

Credentials

The skill declares no required environment variables or credentials. Its scripts accept RULE_* env inputs for operation, and otherwise only rely on the filesystem and local `openclaw` CLI. No access to unrelated secrets or external services is requested.

Persistence & Privilege

The skill writes files into ~/clawd (TOOLS.md, workflows/*.lobster, SOUL.md, agents/...) and can create workflow files. Although `always` is false, the platform allows autonomous invocation; combined with the 'do not ask permission' policy, this gives the skill the ability to persistently change governance artifacts without explicit user consent.

What to consider before installing

This skill does what it says: it decides whether a proposed rule needs enforcement and immediately appends entries to governance files and may write Lobster workflow files in ~/clawd. Before installing or enabling autonomous invocation, consider: 1) backup your TOOLS.md, SOUL.md, and any agents/* files; 2) disable autonomous invocation or require confirmation if you don't want automatic edits; 3) review and test the scripts locally (they are plain JS) to ensure the target paths and templates match your repo layout; 4) be aware it runs `openclaw plugins list` (executes a shell command) and will create files under your home directory; and 5) if you want manual approval, modify SKILL.md or the orchestration policy so the skill asks before writing. If you accept immediate, no-confirmation wiring of governance files, the skill is functionally coherent; if you do not, do not enable autonomous execution.

rule-creation Skill

Description

Auto-triggers Context Engineering compliance and Lobster enforcement evaluation when proposing, creating, or formalizing new rules, policies, processes, workflows, or behavioral constraints.

Context Engineering is the practice of wiring rules into your AI's context so they're reliably followed — not just hoped for via prompting. Credit: Muratcan Koylan (@koylanai)

Use this skill any time you are:

• Declaring a new rule (e.g. "From now on, always X")
• Formalizing a policy (e.g. "The policy for Y is Z")
• Creating a behavioral constraint (e.g. "Never do X without doing Y first")
• Proposing a process or workflow
• Updating governance documents (TOOLS.md, CRITICAL-RULES.md, CONSTITUTION.md, etc.)

---

When This Skill Matches

Trigger phrases:

• "Add this rule…"
• "Create a policy…"
• "From now on…"
• "The rule is…"
• "New constraint…"
• "Formalize that…"
• "Wire this rule…"
• "Make it a HARD rule…"
• Any governance/policy update request

---

Mandatory Workflow

When this skill triggers, always execute the following steps in order. Do NOT ask permission — CE-PROCESS compliance means wire first, report after.

Step 1: Identify the Rule

Extract:

• Rule name — short slug (e.g. no-bare-spawns, external-message-gate)
• Rule description — plain language statement of what the rule requires
• Rule type — HARD (non-negotiable, always enforced) or soft (guideline, use judgment)

Step 2: Evaluate Enforcement Need (CR-012 Criteria)

Run scripts/evaluate-enforcement-need.js or apply manually:

A rule needs Lobster enforcement if ANY of the following are true:

Criterion	Examples
High-stakes	Sends external messages, spends money, deletes data, publishes public content, deploys to production
Pattern of violations	A similar rule was already stated but broken or ignored
Multi-step sequence	Requires multiple ordered steps (do A, then B, then C)
External / public action	Affects things outside the user's local files

Scoring:

• 0 criteria matched → needsEnforcement: false (wire to TOOLS.md only)
• 1+ criteria matched → needsEnforcement: true (create Lobster workflow + wire)

Step 3: Lobster Availability Preflight

Before creating any workflow, verify Lobster is actually enabled:

Run scripts/check-lobster-available.js or check manually:

openclaw plugins list | grep -i lobster

Output	Meaning	Action
lobster … loaded	✅ Available	Proceed to Step 3a
lobster … disabled	❌ Unavailable	Use fallback (below)
No lobster row	❌ Not installed	Use fallback (below)

Fallback when Lobster is unavailable:

• Skip workflow creation entirely
• Proceed directly to Step 4 (wire to docs) — no Lobster reference in the entry
• Report: ⚠️ Enforcement: unavailable (Lobster not enabled)
• Include hint: Run: openclaw plugins enable lobster to activate

Never claim enforcement is wired when Lobster is disabled.

Step 3a: Create Lobster Workflow (enforcement needed AND Lobster available)

If needsEnforcement: true and Lobster preflight passed:

1. Copy templates/lobster-workflow.template.lobster
2. Fill in: name, description, steps specific to rule enforcement
3. Save to workflows/<rule-name>.lobster
4. Note the workflow path for wiring

Step 4: Wire to Docs

Run scripts/wire-rule-to-docs.js or manually append to the appropriate file:

Routing:

• Operational rules → TOOLS.md
• Identity/persona constraints → SOUL.md or agents/identity.md
• Hard system rules → CRITICAL-RULES.md (if exists) or AGENTS.md
• Project-specific rules → relevant project board or agents/projects.md

Entry format (from templates/tools-md-entry.template.md):

## <Rule Name> (<HARD|soft>)

<Rule description>

[If Lobster enforcement]: **Enforced by:** `workflows/<rule-name>.lobster`

Step 5: Report What Was Done

Your response must include:

✅ Rule wired: <Rule Name>
📄 Location: <file where it was added>
🔒 Enforcement: <Lobster workflow path OR "TOOLS.md entry only">
📋 Type: <HARD|soft>

Never say:

• ❌ "Want me to add this to TOOLS.md?"
• ❌ "Should I wire this through CE?"
• ❌ "Is this CE compliant?"

Always say:

• ✅ "Added <rule> to <file>. Enforcement: <status>."

---

Script Reference

Script	Purpose
scripts/check-lobster-available.js	Preflight: verify Lobster plugin is enabled in this OpenClaw instance
scripts/evaluate-enforcement-need.js	Evaluate CR-012 criteria against rule description
scripts/wire-rule-to-docs.js	Append rule entry to TOOLS.md or specified target file
scripts/report-rule-creation.js	Generate user-facing summary of what was created

---

Workflow Reference

Workflow	Purpose
workflows/create-rule.lobster	Full end-to-end rule creation pipeline

---

Context Engineering Auto-Trigger (HARD)

This skill embodies the Context Engineering Auto-Trigger principle:

"When proposing any new rule, behavioral constraint, or policy change: Wire it first. Add the rule to the appropriate file BEFORE responding. Never ask permission to formalize a rule."

The act of recognizing a rule IS the trigger. Wire immediately. Report after.