Skip to content
Back to skills

Ca Conflict

ASecurity

Stop everything and surface a rule conflict — persona vs. docs vs. code. Present both sides and the conflict-hierarchy level; the user resolves. No silent reconciliation.

  • 147 stars
  • 0 votes
  • 0 copies
  • 0 views
  • Added September 2, 2026
ai-agents

Security analysis

A100/100

Scanned September 2, 2026

npx -y skills add arbiterForge/codeArbiter --skill ca-conflict --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Ca Conflict?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Ca Conflict
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/arbiterforge-ca-conflict/badge)](https://www.skillsdirectory.com/skills/arbiterforge-ca-conflict)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: ca-conflict
description: Stop everything and surface a rule conflict — persona vs. docs vs. code. Present both sides and the conflict-hierarchy level; the user resolves. No silent reconciliation.
argument-hint: (none)
---

# $ca-conflict — conflict protocol

Orchestrator protocol, not a skill route. When a rule conflict surfaces — the persona, a `.codearbiter/` document, and code contradict each other, or two docs contradict each other — all other work STOPs and the conflict is presented for the user to resolve. The orchestrator never picks a side.

## Flow

1. **Halt** — suspend the in-progress task. No partial progress.
2. **Identify** the conflicting sources (A and B): which document, file, or rule each is.
3. **Quote** the exact passages side by side.
4. **Classify** the conflict against the §2 conflict-resolution hierarchy and name the level the
   tension sits at. Note which source was more recently updated — informational, not determinative.
5. **Present** and wait. Work resumes only after the user explicitly resolves.

## Output

```
## Conflict detected — work halted

### Source A
File: <path> · Last updated: <date>
> <exact quoted text>

### Source B
File: <path> · Last updated: <date>
> <exact quoted text>

### Nature of conflict
<what the two sources disagree about>

### Conflict-hierarchy level
Level N — <why the tension sits here>

### To resolve
<options — update A, update B, or supersede one via an ADR>

---
Work is halted. Resolve before the orchestrator proceeds.
```

## When routed automatically

The orchestrator surfaces a conflict on its own when the persona and a `.codearbiter/` doc contradict
each other, two docs contradict, code contradicts an accepted ADR, a new ADR contradicts an accepted
one, or a task instruction contradicts a hard gate.

## After resolution

The orchestrator records the resolution; if the user authorized a doc update, it makes that update
(the one sanctioned side effect) and resumes from the suspension point.

## Hard gate

MUST stop ALL other work. MUST NOT pick a side without explicit user instruction. MUST NOT decide by
recency alone. MUST NOT silently continue past a detected conflict. If the conflict involves a guessed
or auto-resolved `[CONFIRM-NN]`, flag that as a separate critical finding.

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.