mismagent-write-code-rules

Writes project-level code-rules.md with enforcement channels for each deliberated coding rule.

Updated Jun 12, 2026
One-click install
npx skills add https://github.com/lucolucus/mismagent --skill mismagent-write-code-rules-lucolucus
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: mismagent-write-code-rules
Source: https://github.com/lucolucus/mismagent/tree/main/codex/skills/mismagent-write-code-rules
Command: npx skills add https://github.com/lucolucus/mismagent --skill mismagent-write-code-rules-lucolucus

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Coding standards often decay into unread documents because nothing enforces them. This Skill writes the project's code-writing rules as a user-visible file where every rule carries an explicit enforcement channel, so a rule without a channel is never written. ## Core Features & Use Cases - Enforcement-channel binding: Each rule is classified as mechanical (gate dependency/style lint such as Konsist, ArchUnit, dependency-cruiser, or import-linter), discursive (code-review criterion), or structural (citation of the owning skill or gate). - Project-level output: Produces <output_dir>/code-rules.md next to profile.md and architecture.md, with the profile's code_rules binding updated so workers and code review receive it automatically. - Delta-based updates: On later features it proposes only deltas against the existing file, keeping one source of truth. - Use Case: During the mismAgent model movement, after the user deliberates rules inside ARCH_PROPOSAL and the stack ADR is set, the architect invokes this Skill to emit code-rules.md with the lint tool and config path pinned per rule. ## Quick Start Ask the mismAgent architect to write the project's code-rules.md from the rules deliberated in ARCH_PROPOSAL after the stack ADR is decided.

Frequently Asked Questions about mismagent-write-code-rules

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I write enforceable coding rules for a project?

Write each rule together with its enforcement channel: a gate lint for mechanical rules, a code-review criterion for discursive ones, or a structural citation of the owning skill. A rule with no channel is not written, since unenforced rules decay.

What dependency lint tools enforce architecture rules per stack?

JVM/Kotlin uses Konsist or ArchUnit, TypeScript uses dependency-cruiser or eslint-plugin-boundaries, and Python uses import-linter. The lint config lives in the side's repo, wired at wave-0 scaffold, and runs in the worker's gate loop, verifier step 2, and CI.

Where does the code-rules.md file live in a mismAgent project?

It is project-level, placed in the output directory next to profile.md and architecture.md, not per-feature. The profile points at it through the code_rules binding, so workers and code review receive it through the profile they already read.

When is the write-code-rules skill invoked in the mismAgent flow?

It is invoked by mismagent-architect in pass-2 of the model movement, after the user deliberated the rules inside the ARCH_PROPOSAL checkpoint and after the stack ADR, since the lint choice depends on the selected stack.

Why not use grep to enforce dependency rules?

A dependency rule is a graph property, and POSIX grep is blind to fully-qualified-name use, build files, and aliases. The stack's dependency lint tool evaluates the real import graph, so its config is derived from architecture.md's module map.