AGENTS.md@plugins/preset-cli-skills git:20260730.ca70ab8A

Immutable. This exact content is served forever at /api/v1/blob/4ca8c1e0bbfe34ee.

# Preset CLI Skills

This package contains installable Preset CLI skills for OpenAI Codex, Gemini CLI, and direct repository readers.

Use these skills only for explicit Preset CLI work driven through the `sup` CLI (PyPI package `superset-sup`). If the user is working through Preset/Superset MCP tools, use the separate `preset-mcp-skills` package instead. If the user wants direct HTTP, SDK, or `requests`/`curl` code paths, use the separate `preset-api-skills` package instead. Do not switch surfaces unless the user explicitly approves the switch.

## Surface Selection

- If the user mentions MCP, MCP tools, MCP clients, Superset MCP, Preset MCP, or Copilot/MCP behavior, do not use this package. Route to `preset-mcp-skills`.
- If the user asks for in-process HTTP, REST endpoints, Python `requests`, SDKs, or curl examples, do not use this package. Route to `preset-api-skills`.
- Use this package only when the user asks for `sup`, the Preset CLI, shell one-liners, scripting, batch exports, ad-hoc SQL from a terminal, or CI/CD automation that is simpler as a single command than as an HTTP call.
- If both API and MCP plugins are installed, MCP intent wins over resource type. A dashboard, chart, dataset, or SQL Lab request should still use MCP guidance when the user asked for MCP.
- If a CLI workflow lacks the needed capability, stop and ask whether to switch to the API surface. Do not silently escalate.

## Skill Routing

Use the `skills/*/SKILL.md` files as the canonical instructions:

- `skills/preset-cli/SKILL.md` - drive the `sup` CLI for non-destructive shell, scripting, and CI/CD workflows.
- `skills/preset-cli-mutations/SKILL.md` - state-changing `sup` CLI operations with mandatory preview and confirmation templates.

Detailed examples live in each skill's `references/` directory. Load only the reference files needed for the user's task.

## Client Entry Points

- OpenAI Codex: `.codex-plugin/plugin.json` plus this `AGENTS.md`.
- Claude Code: `.claude-plugin/plugin.json` plus `skills/*/SKILL.md`; Claude plugin installs do not load package-level `AGENTS.md` or `CLAUDE.md` context.
- Gemini CLI: `skills/*/SKILL.md` (this `AGENTS.md` is only for legacy context imports).
- Direct repository readers: this `AGENTS.md`.
- Cursor: `.cursor-plugin/plugin.json`.
- GitHub Copilot: `.github/copilot-instructions.md`.

## Safety Policy

Default to non-destructive reads and exports. Before any state-changing `sup` command — `sup chart push`, `sup dashboard push`, `sup dataset push`, `sup user push`, `sup user invite`, any `--force` or `--overwrite` invocation, or `sup sync run` against any target workspace — preview the change first (native `--dry-run` for `sup sync run`, `sup user push`, and `sup user invite`; pull-and-diff against the target for chart/dashboard/dataset push which have no native `--dry-run`), summarize the source workspace, target workspace, asset IDs/types, and any destructive flags, then get explicit user confirmation that names the target workspace and the literal flag strings.

Never paste `SUP_PRESET_API_TOKEN`, `SUP_PRESET_API_SECRET`, or any bearer token into a command line; rely on environment variables and `sup config auth`. Redact tokens, refresh tokens, database passwords, and any credential surfaced in `sup` output before sharing transcripts, logs, or screenshots.