Skills Overview
Official agent skills for the Copera platform
Copera Skills are portable instruction bundles that teach AI coding agents how to use the copera CLI on a real workspace. They make agents fluent in Copera commands — and let you capture your team's specific workflows so an agent can repeat them reliably.
One source, multiple install paths — see Install for the command for your agent of choice.
Two tiers
Built-in. Teaches the agent every copera command, the JSON shapes it returns, the discovery-before-write rule, exit-code semantics, and rate-limit behavior. Ships with this repository, works on any workspace.
User-generated. Capture your workspace structure (this board, that table, those columns) plus a step-by-step procedure ("triage a bug", "send a weekly report"). Built by interviewing you and probing your workspace. Per-script confirmation flags + reactive schema-drift detection on copera errors.
The CLI skill is the foundation; workflow skills sit on top of it. You install the CLI skill once, then ask any agent that loaded it to build a workflow skill for any procedure you do repeatedly.
Why skills exist
Copera is intentionally unopinionated — no two workspaces look alike. A bare LLM agent given the CLI cannot effectively act on your behalf because:
- It doesn't know which board is "the bug tracker" vs "the CRM".
- It doesn't know that column
col_a4f2means "Status" with valuesTriage / In Progress / Done. - It doesn't know your conventions ("every new bug gets a doc under
/Engineering/Bugs/"). - Re-discovering all of that costs tokens every single conversation.
Skills are how that workspace knowledge gets captured, versioned, and reused — without leaking secrets, without locking you into a vendor format, without duplicating instructions across tools.
What's inside the CLI skill
A single router (SKILL.md) plus on-demand reference files:
| When the user asks about… | The agent reads |
|---|---|
| Logging in, tokens, profiles | reference/auth-and-profiles.md |
| Boards, tables, column schemas | reference/boards-tables.md |
| Reading, creating, updating rows | reference/rows.md |
| Markdown documents | reference/docs.md |
| Drive files and folders | reference/drive.md |
| Row comments and channel messages | reference/comments-channels.md |
| Errors, exit codes, rate limits | reference/errors-and-retries.md |
| Building a custom workflow skill | workflows/build-workflow-skill.md |
This progressive disclosure keeps the agent's base context small — only the relevant reference loads when needed.
Supported agents
| Agent | How |
|---|---|
| Claude Code | Native plugin via /plugin marketplace add |
| Cursor | Native plugin via /add-plugin |
| OpenCode | Reads .claude/skills/ natively, or use npx skills add |
| Codex CLI | Universal install via npx skills add |
| Windsurf, Cline, Roo, Aider, Continue, others | Universal install via npx skills add |
The same SKILL.md works in every agent — npx skills auto-detects which agents are installed and writes to the right path. See Install for copy-paste commands per agent.
License
Copera Skills are MIT-licensed and developed in the open at github.com/copera-ai/copera-skills. Contributions welcome.