@oussamadouhou/agent-enforcement

Evidence enforcement hooks for OpenCode and Claude Code. This package blocks or warns when todowrite tasks are marked complete without the required evidence block, checklist, and trust markers.

Installation

bun add @oussamadouhou/agent-enforcement

OpenCode usage

import type { Plugin } from "@opencode-ai/plugin"
import { createEvidenceEnforcementHook } from "@oussamadouhou/agent-enforcement"

const AgentEnforcementPlugin: Plugin = async ctx => {
  return createEvidenceEnforcementHook(ctx)
}

export default AgentEnforcementPlugin

Claude Code usage

Copy the plugin to your Claude Code plugins directory and enable hooks/enforcement.py:

cp -r . ~/.claude/plugins/agent-enforcement

Then ensure hooks/hooks.json is configured in your Claude settings.

Configuration

• OPENCODE_ENFORCEMENT_LEVEL or ENFORCEMENT_LEVEL: 0 (creative), 1 (standard), 2 (strict)
• ENFORCEMENT_LOG_MODE: file | console | silent
• ENFORCEMENT_LOG_PATH: file path for logs
• ENFORCEMENT_LOG_LEVEL: debug | info | warn | error

Evidence format

**Evidence for task-1**:

**Execution**:
- Tool: read

**Verification**:
- Checked output

**Checklist**:
- [x] Tool executed successfully
- [x] Output captured
- [x] Result matches expected
- [x] No workarounds used
- [x] Independently verifiable

**Trust**: 🟢 HIGH | ✅ Ground Truth

Development

bun test
bun run build