2.5k
GitHub Stars
2
Bundled Files
2 months ago
Catalog Refreshed
4 months ago
First Indexed
Readme & install
Copy the install command, review bundled files from the catalogue, and read any extended description pulled from the listing source.
Installation
Preview and clipboard use veilstrat where the catalogue uses aiagentskills.
npx veilstrat add skill openclaw/skills --skill opencode-omo- _meta.json462 B
- SKILL.md2.7 KB
Overview
This skill documents how to use OpenCode together with the Oh-My-OpenCode operator, prioritizing a Sisyphus-first workflow that leverages Prometheus for planning and Atlas for execution. It describes commands, core rules, and failure-handling patterns to make iterative coding safe, repeatable, and auditable. The guidance emphasizes interacting through OpenCode rather than editing code outside the platform.
How this skill works
OpenCode acts as the primary environment for planning and writing code. Use the Sisyphus agent for one-shot or interactive sessions, call Prometheus via @plan to generate structured plans, and trigger Atlas with /start-work to execute plans. One-shot runs use the opencode CLI with the sisyphus agent and ultrawork (ulw) to execute small, explicit requests deterministically.
When to use it
- When you want a repeatable, auditable coding workflow within OpenCode.
- For complex tasks that require planning followed by automated execution.
- When you need to avoid direct edits outside the OpenCode environment.
- For quick, deterministic one-shot changes using ulw.
- When you need interactive clarification during implementation.
Best practices
- Always prefer Sisyphus-first: start with the sisyphus agent for planning and iteration.
- Use @plan in Prometheus to capture decisions, constraints, and edge cases before execution.
- Trigger /start-work in Atlas only after the plan is approved and clarified.
- Keep one-shot ulw requests small and explicit to improve determinism.
- Answer agent clarifying questions in plan mode and re-run execution rather than editing code externally.
Example use cases
- Plan and implement a multi-file refactor: generate the plan with @plan, review, then run /start-work.
- Apply a targeted bug fix with a single deterministic ulw command via opencode run --agent sisyphus "ulw <request>".
- Audit and archive skill versions by running tasks that read, document, and store changes through OpenCode workflows.
- Iteratively build a feature: plan with Prometheus, refine in Sisyphus, and execute incremental steps with Atlas.
FAQ
Check the OpenCode config file and look for "oh-my-opencode" in the plugin list, for example by printing ~/.config/opencode/opencode.json.
What if the agent asks clarifying questions during execution?
Answer those questions in plan mode (Prometheus), update the plan, and re-run the execution with Atlas rather than editing files outside OpenCode.