Agent config adapter
Linlin's curated AI agent harness configuration: workflow rules, skills, hooks, plugin recommendations, tooling preferences, and project templates. Loadable into any new project so a fresh /init can pick the relevant subset.
npx -y skills add jajupmochi/agent-harness --skill agent-config-adapterAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Adapt an existing agent configuration or plugin to a new agent or model route. Use when moving agent-harness or another agent setup between Claude Code, Codex, Gemini, Cursor, local models, or non-native model backends such as DeepSeek routed through another agent.
SKILL.md
2.6 KB, as published. Nobody here has run it
agent-config-adapter
Use this workflow to port an existing agent configuration without dumping every rule into the target agent context.
Inputs to establish
- Source agent and config root.
- Target agent and target model route.
- Whether the model is native to the target agent or routed through a compatibility layer.
- Required capabilities: durable instructions, skills, hooks, MCP/tools, slash commands, subagents, browser/computer use, memories, and installers.
- Non-negotiable isolation constraints: what must not affect the original agent setup.
Adaptation steps
- Inventory the source configuration:
- manifests
- skills
- hooks
- rules/instructions
- MCP/app/tool config
- install scripts
- templates
- tests or validation scripts
- Research the target agent's current extension surfaces from local docs or
official docs. For Codex, use the
openai-docsCodex manual route. - Build a mapping table: source item, target surface, required rewrite, verification method, and isolation risk.
- Choose the smallest target entrypoint:
- instructions file for always-on repo rules
- skill for reusable workflow
- plugin for distribution
- hook for lifecycle enforcement
- MCP/app for live tools or private external data
- Implement target-specific wrappers. Keep shared source content as references so implicit skill metadata stays small.
- Verify structurally first, then run one realistic prompt per major workflow.
Model-route fallbacks
When the target model is not native to the agent, or implicit tool use is weak:
- Prefer explicit skill invocation in user docs and default prompts.
- Split long rule sets into small wrappers that name exactly which references to read.
- Use scripts for deterministic checks instead of relying on the model to remember every invariant.
- Require command output or file inspection before success claims.
- Avoid hidden global instructions that the routed model may ignore.
- Keep model-specific workarounds in the adapter skill, not in shared rules.
Deliverables
Every adaptation should leave:
- A plan document with the mapping table and selected architecture.
- Agent-specific manifest/config files.
- A validation command or script.
- Installation notes for the target agent.
- A rollback note explaining which original agent files were not touched.