| name | pluxx-translate-hosts |
| description | Review preserve, translate, degrade, and drop behavior across Claude Code, Cursor, Codex, and OpenCode. |
Pluxx Translate Hosts
Use this skill when the user needs to understand how a Pluxx source project maps into the core four.
The goal is not to promise fake parity. The goal is to preserve author intent and explain where each host gets a native surface, a translated equivalent, a weaker equivalent, or no honest equivalent.
Workflow
- Run or inspect current validation output:
pluxx lint
pluxx build --target claude-code cursor codex opencode
- Review the highest-risk surface:
- commands and argument UX
- specialist agents/subagents
- hooks and trust behavior
- permissions and tool approval intent
- MCP, auth, and runtime materialization
- Explain the mapping by bucket:
- instructions
- skills
- commands
- agents
- hooks
- permissions
- runtime
- distribution
- For every important caveat, classify it as:
- preserve
- translate
- degrade
- drop
- Recommend the best source-shape fix when a degraded host could be improved by moving intent to commands, agents, permissions, hooks, or runtime config.
Rules
- Do not flatten host-specific nuance into “supported everywhere.”
- Do not hand-edit generated
dist/ as the fix; fix the source project.
- Prefer improving source authoring shape over adding target-specific hacks.
Output
- concise preserve/translate/degrade/drop matrix
- highest-value source-shape fix
- target-specific caveats users need to know before publishing