Setting the issue. One moment.
Contents
Repo No. 44Dreambase · Updated 17 September 2026 · In print
Skills Docs · No. 44
9 skills9,544 words
Inside this issue
1 — Install
npx skills add DreambaseAI/skills/plugin marketplace add DreambaseAI/skillsMIT4 stars9,544 wordsUpdated 17 September 2026
2 — Skills
9 skills · 42 minutes- Dreambase Data PresentationDesign and build data presentations — slide decks, board and investor decks, QBRs, analytics readouts, launch…8 min
- Dreambase Data StoriesDesign best-in-class data-story artifacts — reports, executive summaries, slide decks, one-pagers…3 min
- Dreambase EChartsAuthor Apache ECharts option (EChartsOption) JSON configs for the Dreambase renderer.2 min
- Dreambase Industrial SchematicsCreate dark, cinematic technical-schematic illustrations and UI components in the cinematic industrial / HUD…5 min
- Dreambase MCPUse Dreambase workspaces, dashboards, datasets, the durable-dataset lifecycle, connected Supabase/API/MCP…3 min
Skills Docs · Repo No. 44Whole repo as markdown
Provenance
- Owner
- DreambaseAI · Organization
- Licence
- MIT
- Last updated
- Published from
- github.com/
DreambaseAI/ skills
Typeset in this issue’s own colours, resolved from the owner’s published design.md.
Front matter
From the READMEDreambase
The dreambase plugin: the Dreambase MCP server plus the agent skills for
querying it and turning the results into charts, reports, and decks.
Install the plugin
Claude Code
# One time: register this repository as the Dreambase plugin marketplace.
/plugin marketplace add DreambaseAI/skills
# Install the complete Dreambase plugin from that marketplace: all skills and the MCP server.
/plugin install dreambase@dreambaseCursor — add this repo as a plugin marketplace in Cursor’s plugin settings,
then install dreambase.
Codex — point Codex at this repo’s marketplace and install dreambase.
# One time: register this repository as the Dreambase plugin marketplace.
codex plugin marketplace add DreambaseAI/skills
# Install the complete Dreambase plugin from that marketplace: all skills and the MCP server.
codex plugin add dreambase@dreambaseNo separate skill or MCP installation is required when using the plugin.
The plugin declares the MCP server as a remote type: "http" connection, so your
host runs the OAuth sign-in itself the first time a Dreambase tool is called — no
local process, no token to paste. Restart your client after installing, then run
whoami followed by list_workspaces to confirm it works.
New to Dreambase? SETUP.md (opens in a new tab)
walks through creating an account, connecting a Supabase project, and completing
consent.
What the MCP server does
https://app.dreambase.com/mcp exposes your Dreambase workspace to an agent:
| Area | Tools |
|---|---|
| Identity | whoami, list_workspaces |
| Dashboards | list_dashboards |
| Datasets | list_datasets, get_dataset, query_dataset, plan_datasets, save_dataset, refresh_dataset, promote_dataset |
| Metric snapshots | list_aggregates, get_aggregate |
| Connections | list_connections, get_connection, search_connection |
| Connector setup | list_connectors, request_connector_connection, get_connection_request |
| Database health | create_health_report, list_health_reports, get_health_report |
| Workspace Skills | list_skills, get_skill, create_skill, update_skill |
query_dataset runs a sandboxed read-only DuckDB SELECT over one dataset, so
an agent can compute an answer instead of paging through rows. Datasets start as
scratch snapshots that expire in about a week; once you’ve verified one,
promote_dataset keeps it permanently, and refresh_dataset replays its saved
definition for current rows. Access is gated by
OAuth scopes and your workspace membership: the tool list you see is exactly
what your grant allows. The dreambase-mcp skill teaches an agent how to drive
all of it correctly.
The skills
All nine canonical skills ship with the plugin and are also installable on their own. This preserves the complete root skillset while adding plugin-based installation.
| Skill | What it does |
|---|---|
dreambase-mcp (opens in a new tab) | The usage contract for the Dreambase MCP server — call order, the query_dataset DuckDB dialect and its result caps, connection discovery, async health-report polling, retry rules for non-idempotent writes, the durable-dataset lifecycle (save → verify → promote) and its consent gate, and what each error code means |
dreambase-echarts (opens in a new tab) | Author Apache ECharts option JSON for the Dreambase renderer — includes an offline option-schema lookup CLI and a structural config validator |
dreambase-visualization-design (opens in a new tab) | Design, critique, and improve charts, dashboards, and infographics using evidence-based visualization, perceptual, accessibility, and integrity principles |
dreambase-data-stories (opens in a new tab) | Design reports, executive summaries, slide decks, infographics, and scrollytelling pieces that blend charts, illustration, typography, and narrative — built on the practices of FT, NYT Graphics, The Pudding, and Reuters |
dreambase-data-presentation (opens in a new tab) | Design and build data presentations — board, investor, QBR, readout, launch, sales, and keynote decks — as interactive self-contained web decks, native .pptx, Google Slides, or a build spec, with verified slide-craft specs, staged chart reveals, and presenter mechanics |
dreambase-public-reports (opens in a new tab) | Create disclosure-safe public artifacts from real or sensitive data using transformed visuals, constructive and honest storytelling, and explicit approval for visible percentages, rates, deltas, ratios, or indexes |
dreambase-industrial-schematics (opens in a new tab) | Create dark cinematic industrial/HUD technical schematics — orbital process rings, cel-shaded exploded machinery, holographic 2D/3D wireframe mesh charts with monochromatic hue themes, tick-strip components — as portable SVG/HTML, with zero-dep ring/ruler and 3D-mesh generators |
dreambase-micrographic-design (opens in a new tab) | Design in the micrographics style — compliance-labelling visual language (rating plates, care tags, spec sheets) used as intentional design: hairline rules, boxed compartments, tick scales, LABEL: value pairs, mark clusters, micro-typography, monochrome plus one safety accent — with the certification-mark legal guardrail |
dreambase-skill-creator (opens in a new tab) | Create, improve, validate, and evaluate Dreambase agent skills using this repository’s conventions and eval loop |
How they compose: dreambase-mcp gets the data out.
dreambase-data-stories shapes the narrative artifact,
dreambase-visualization-design decides each chart, and dreambase-echarts
builds and validates the rendered config. When the artifact is a deck,
dreambase-data-presentation takes the storyline and owns the slide, the room,
and the presenter. Add dreambase-public-reports when the artifact uses
confidential real data but must communicate honest trends without publishing the
underlying values.
Installing a skill without the plugin
# Interactive — pick from the repo's skills
npx skills add DreambaseAI/skills
# See what's available without installing
npx skills add DreambaseAI/skills --list
# Install specific skills by name (repeat --skill for several)
npx skills add DreambaseAI/skills --skill dreambase-mcp --skill dreambase-echarts
# Or point at one skill's directory directly
npx skills add https://github.com/DreambaseAI/skills/tree/main/skills/dreambase-echarts
# Everything, no prompts
npx skills add DreambaseAI/skills --allOr copy a directory into your skills folder: cp -r skills/dreambase-<name> ~/.claude/skills/
(per-user), or into <your-project>/.claude/skills/ (per-project).
Repository structure
.agents/plugins/marketplace.json # Codex marketplace
.claude-plugin/
├── marketplace.json # Claude marketplace
└── plugin.json # Claude plugin manifest
.cursor-plugin/
├── marketplace.json # Cursor marketplace
└── plugin.json # Cursor plugin manifest
.codex-plugin/plugin.json # Codex plugin manifest
.mcp.json # Remote HTTP MCP server (Claude/Codex)
mcp.json # Remote HTTP MCP server (Cursor)
assets/logo.png # 1024x1024 brand mark
SETUP.md # Guided first-run setup
skills/dreambase-<name>/ # Canonical skill and plugin component
├── SKILL.md # Required: frontmatter + instructions
├── scripts/ # Optional: executable helpers
├── references/ # Optional: docs loaded on demand
├── assets/ # Optional: templates, files used in output
└── evals/evals.json # Test prompts + assertions (committed)
scripts/build-plugin.mjs # Validate and build store artifacts
store/main-skills.json # Cannot-delete baseline for pre-release skills
dist/dreambase/ # Generated, symlink-free store artifactThis is a single-plugin repository, so the repository root is also the plugin
root. skills/ is both the canonical source and the standard auto-discovered
plugin component directory—there is no copied or symlinked second tree to
drift. pnpm check:plugin verifies the manifests and guarantees every skill in
the pre-release main inventory remains present. pnpm build:plugin creates a
self-contained dist/dreambase/ submission artifact.
Contributing
Read CONTRIBUTING.md (opens in a new tab). The short version: open this repo in
Claude Code and describe the skill you want — the dreambase-skill-creator
meta-skill triggers automatically and walks the full loop (interview → draft →
test → evaluate → iterate). Validate before committing:
node skills/dreambase-skill-creator/scripts/validate-skill.mjs --allConventions
- Naming: directory name = frontmatter
name=dreambase-<kebab-case>. - SKILL.md stays under ~500 lines; push detail into
references/. - Descriptions state what the skill does and when to trigger it (be generous with trigger phrases — models under-trigger).
- Eval prompts live in
evals/evals.jsonand are committed; run outputs (*-workspace/directories) are gitignored. - Scripts are Node-first: zero-dependency
.mjs(Node built-ins only) or POSIX shell. We’re a Node/Next.js shop and Node is guaranteed everywhere our skills run.
License
Colophon
How this issue was made- Source
- github.com/DreambaseAI/skills, branch
main, last pushed 17 September 2026. - Licence
- MIT — the text of every skill is reproduced unmodified, frontmatter included, under the upstream licence.
- Discovery
- 9 skills found by walking the repository tree for
SKILL.md, not by matching a directory convention. One layout observed:skills/*/SKILL.md. - Issue colours
- Resolved from the owner’s published design.md — hue 151°, chroma 0.190. Two accent tones are generated per issue and each is proven against its own ground before it ships: a single accent that passes AA on both light and dark paper is arithmetically impossible.
- Heading repairs
- 1 repair applied to the front matter so the document has one
h1and no skipped levels:- Shifted “9 headings” from h1 to h2 so the skill title is the only h1.
- Spec compliance
- Every skill satisfies the Agent Skills specification’s frontmatter rules.
- Images
- Images inside a skill come from the upstream repository. Where the author gave no alternative text we mark the image decorative rather than inventing a description — a plausible caption we made up is worse than none for the reader who depends on it.
- Marketplace
- A plugin manifest is published at
.claude-plugin/marketplace.jsonby Dreambase, declaring 1 plugin. It is read for editorial metadata only — never as the skill index, which is always the repository tree. - Signal
- This repository is unranked on skills.sh, which says nothing about it: the leaderboard covers a fraction of published skills.
- Agent surfaces
- The whole issue is available as one markdown document at
/DreambaseAI/skills.md, and each skill at its own.mdURL. - Publication
- Set by Skills Docs from the source repository. Body text is Literata at the reader’s chosen size and measure; code is Geist Mono. Nothing on this page was written by us except this paragraph.