SandBase Skills Library

Installable open-source Agent Skills for research, social intelligence, marketing and business workflows, working with Claude Code, Codex, Cursor, Gemini CLI and DeepSeek Harness.

Stars
★ 212
Last updated
1mo ago
License
Apache-2.0
Primary language
Python

At a glance

How it runs
Agent plugin / skillCLI
Works with
Universal · cross-platformCodex · Claude CodeChatGPT (Needs adaptation)
Cost
Free software; you pay for model usage
Setup effort
Low · running in minutes
You'll need
Node.js (npx)Python 3 (validator)GitHub CLI 2.90.0+ (gh skill)SANDBASE_API_KEY (optional, for specialized social/market/data Skills)Shell / CLINetwork accessLocal filesystemMCP Server
Typical use
A market or intelligence analyst verifying a market claim, using multi-source-search to cross-check independent sources and emit an auditable evidence ledger
Not a fit if
  • Users who want a single chat assistant and won't install instruction files into an AI agent
  • Teams that need deep social/market provider capabilities but refuse to register a SandBase account
Website
skills.sh ↗ set on GitHub
Maintenance
Commits in the last 6 monthslast commit 2026-09-26latest release v0.3.5 (2026-08-20)
Source review
65/100 · Some gaps

What does this agent do, and when should you use it?

SandBase Skills (repository sandbaseai/sandbase-skills) is an Apache-2.0 open-source collection of installable Agent Skills — the README states 245 (the repo description says 88) — organized into research, social intelligence, business intelligence, marketing & SEO, media generation, data science, software engineering, and utility catalogs. Each Skill is a SKILL.md instruction file that defines a repeatable workflow, evidence rules, and an output format that a host agent reads and executes. The repository separates hand-written Skill sources (research/, marketing/) from generated catalog metadata (catalog/skills/, skills.json) and per-client packaging (agent-plugin/, .claude-plugin/, dsh/), with consistency enforced by npm test, marketplace:check, and scripts/skillpack.py validate. The flagship multi-source-search Skill runs on host-provided search and page-reading tools and produces an evidence ledger that can be validated offline, with no SandBase account required; specialized social, market, and data Skills additionally use SANDBASE_API_KEY to tap SandBase's 2,000+ models and APIs. The deployment boundary is clear: Skills are instructions only — execution, searching, and media generation happen in the host agent and whatever tools it is connected to.

Once installed, the agent reads each Skill's SKILL.md and follows the workflow it defines. For example, multi-source-search cross-checks a claim using the host's existing search tools, builds a source-linked evidence ledger, and can verify report consistency offline:bash
python3 research/multi-source-search/scripts/validate_report.py examples/verifiable-research-report.json

` The validator rejects unknown or duplicate sources, inflated confidence, unused evidence, and high-confidence claims with a declared conflict. The 14 social Skills (twitter-intelligence, youtube-research, weibo-research, xiaohongshu-research, …) search posts, track trends, and analyze sentiment across major platforms; market-research and competitor-monitor produce market and competitive intelligence; seo-content-brief generates SERP-backed content briefs. With SANDBASE_API_KEY configured, Skills call SandBase providers via sandbase_discover → sandbase_inspect → sandbase_run (polling async results with sandbase_run_get); the separate SandBase CLI exposes 2,000+ models and APIs to the same agent as six MCP tools.

  1. A market or intelligence analyst verifying a market claim, using multi-source-search to cross-check independent sources and emit an auditable evidence ledger
  2. A social media manager comparing this month's sentiment and recurring complaints for two brands on X via twitter-intelligence
  3. A growth marketer tracking three competitors' pricing, launches, content, and social activity with competitor-monitor's dated competitive-intelligence brief
  4. An SEO content lead producing a writer-ready brief for a target keyword, with search intent, competing pages, and differentiation angles from seo-content-brief
  5. A recruiter or investor assessing an engineering team's open-source activity with github-profile-research's repository, language, contribution, and star analysis
  6. Researchers in the China market monitoring topics and sentiment across WeChat, Weibo, Xiaohongshu, and Douyin using weibo-research, xiaohongshu-research, and china-social-research

Key setup facts

Pulled from the install and usage notes below. The project README has the current values.

Install / run
  • npx skills use sandbaseai/sandbase-skills@multi-source-search
  • npx skills add sandbaseai/sandbase-skills@multi-source-search --agent codex
  • npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz connect --client codex
  • npx skills add sandbaseai/sandbase-skills@<skill-name> --agent codex --global
Environment variables
SANDBASE_API_KEY
Config files
examples/verifiable-research-report.json

How do you install or deploy this agent?

Skills are free and open source (Apache-2.0). Try without installing:
bash

npx skills use sandbaseai/sandbase-skills@multi-source-search

Install into Codex:
bash

npx skills add sandbaseai/sandbase-skills@multi-source-search --agent codex

GitHub CLI 2.90.0+ can preview before installing:
bash

gh skill preview sandbaseai/sandbase-skills research/multi-source-search
gh skill install sandbaseai/sandbase-skills research/multi-source-search --agent codex --scope user

Portable Agent Plugin (Copilot CLI and compatible clients):
bash

copilot plugin install sandbaseai/sandbase-skills:agent-plugin

DeepSeek Harness (all Skills):
bash

dsh plugin --profile web add github:sandbaseai/sandbase-skills
dsh web

Claude Code marketplace (all Skills):
text

/plugin marketplace add sandbaseai/sandbase-skills
/plugin install sandbase-skills@sandbase-agent-skills

The flagship Skill needs no account when the host already provides search tools; specialized social/market/data Skills require SANDBASE_API_KEY in your environment — never in a prompt or committed file.

How do you use this agent?

Install a Skill, then give your agent a task, e.g. to multi-source-search: "Fact-check this claim with independent sources and validate the evidence ledger." The flow: user question → agent reads SKILL.md → host-compatible search or browser tools are used first → if configured, SandBase providers are called via sandbase_discover/sandbase_inspect/sandbase_run → the agent executes the workflow, validates structured evidence, and delivers the result. Verify research output offline:
bash

python3 research/multi-source-search/scripts/validate_report.py examples/verifiable-research-report.json

To add SandBase's 2,000+ models and APIs as MCP tools in the same agent:
bash

npx -y https://github.com/sandbaseai/cli/releases/download/v0.1.17/sandbaseai-cli-0.1.17.tgz connect --client codex

Install other Skills by name:
bash

npx skills add sandbaseai/sandbase-skills@<skill-name> --agent codex --global
npx skills add sandbaseai/sandbase-skills --list

What are this agent's strengths and limitations?

Pros
  • The flagship multi-source-search runs entirely on host-provided tools with an offline evidence-ledger validator — zero accounts and no vendor lock-in
  • One Skill source ships to many agents: Codex, Claude Code marketplace, Copilot CLI Agent Plugin, DeepSeek Harness bundle, plus npx skills add for Cursor, Gemini CLI and more
  • Broad catalog — 245 Skills spanning multi-platform China social research (WeChat, Weibo, Xiaohongshu, Douyin), SEO, competitor monitoring, data science, and software engineering — with catalog metadata and CI consistency checks
  • The evidence-ledger validator concretely rejects duplicate sources, inflated confidence, and unused evidence, making research output auditable
Limitations
  • Skills are instruction files only; real-world quality depends heavily on the host agent and its tools, and the repo does not provide per-Skill verifiable evidence for all 245 entries
  • Specialized social, market, and data Skills require SANDBASE_API_KEY, and SandBase API calls are usage-based (typical research task $0.05–$0.20), creating partial dependence on the SandBase ecosystem
  • The Skill count is inconsistent in the source (88 in the repo description vs. 245 in the README), and many Skill descriptions are templated boilerplate, so depth must be verified per Skill
  • The offline validator checks internal consistency only; it does not attest that a source is true — trust remains with the user

How does this agent compare with similar options?

The repository distributes its Skills through multiple directories (skills.sh, AgentSkill.sh, Agentic Awesome Skills, Agent Skill Exchange, askill, and others) and states compliance with the Agent Skills specification and the vendor-neutral Agent Plugins 1.0 specification; its differentiator versus client-specific skill collections is that the same canonical source is packaged for Codex, Claude Code marketplace, Copilot CLI, and DeepSeek Harness. The source does not provide a functional comparison against specific competitor libraries.

Key facts side by side with the most closely related agents.

Agent Source review Form / cost Stars Updated Language Full support on
SandBase Skills Library This agent 65 · Some gaps Agent plugin / skillFree + model costs ★ 212 1mo ago Python Codex · Claude Code
Material Learning Studio 82 · Good Agent plugin / skillFree ★ 1.5k 1mo ago Python Codex · Claude Code
DeepChat 70 · Some gaps Desktop appFree + model costs ★ 6.4k 1d ago TypeScript OpenAI API · Claude API
Thinking Methods 67 · Some gaps — ★ 25 2d ago Python Codex · Claude Code

How does FollowAgents rate this agent?

FollowAgents source review · FARS-2.1
Some gaps
65/ 100 5-point scale 3.3 / 5
Trust 18/29
Reliability 8/14
Adaptability 14/18
Convention 11/18
Effectiveness 9/13
Verifiability 5/8
Why each dimension lost points
Trust18 / 29 · 3.1/5

The sampled Skills ask agents to send only necessary data, prefer tools the user already has, and obtain confirmation after presenting endpoints, arguments, prices, call counts, and estimates before paid calls. They also prohibit silent provider switching and fabricated results, so external effects and authorization boundaries are addressed. Reading the API map only when needed limits default data flow, and the security policy asks reporters to remove credentials and personal data. Deductions: the sample does not specify local tool permissions, secret storage, or concrete retention controls; installation uses npm install with scripts disabled and audit off, but no dependency lock or vulnerability mitigation is shown; the installer requires force to replace an existing directory, while recovery steps are not described. Source and handoff traceability are encouraged, but item-level citations for external facts are not supplied.

Reliability8 / 14 · 2.9/5

The packet shows automated checks and CI configuration for catalog consistency, report structure, installation, packaging, and plugin manifests, with guidance for unavailable providers, validation failures, and asynchronous runs. This supports ordinary failure handling and dependency availability to a degree. Deductions: README.de.md claims 245 Skills while the candidate description says 88; with only a sample of Skills, consistency with this revision cannot be established. CI configuration is static evidence and does not prove checks passed; detailed failure messages are evidenced mainly through tests of the sample validator.

Adaptability14 / 18 · 3.9/5

The product description covers research, marketing, social intelligence, and business workflows. Sample Skills include trigger descriptions and distinguish local work from tasks that genuinely need external data or generation. The README claims support for several Agents, with dedicated DeepSeek Harness installation instructions. Deductions: the sample covers only some Skills, so scenario boundaries and compatibility across all 88 Skills are unverified; some triggers are broad, and no complete environment capability matrix or limitations are shown.

Convention11 / 18 · 3.1/5

The sample provides categorized directories, installation commands, DeepSeek instructions, a worked research example, and some FAQ or limitation guidance. Package metadata and the license file identify Apache-2.0; the security policy gives a private reporting route and supported-version guidance. Deductions: README.de.md's Skill count conflicts with the candidate description; the supplied materials show no clear changelog, and maintenance responsibility is described only through maintainers and a GitHub report path, without response timing or a named owner. The sample cannot establish that unprovided documentation is absent.

Effectiveness9 / 13 · 3.5/5

Skills provide structured workflows, quality gates, and handoff requirements. The research example emphasizes sources, claims, conflicts, and evidence gaps; the core research Skill says it can use the Agent's existing search tools, with specialized SandBase calls optional, supporting practical value and cost control. Deductions: this sample cannot establish the independent benefit, actual output quality, or relative effort of all 88 Skills; some workflows are generic checklists, and deliverable standards vary by Skill.

Verifiability5 / 8 · 3.1/5

The research workflow calls for separating verified facts from interpretation, retaining sources, and stating uncertainty. The example validator tests duplicate URLs, unused sources, and inflated confidence; the CI file lists catalog, test, and packaging checks. Deductions: no execution results or independent validation are provided, and CI is configuration evidence only; examples and tests cover a small number of research and installation paths, so cross-source corroboration and traceability across other Skills remain uncertain.

Risks and how to mitigate them
  • The Skill count in README.de.md conflicts with the candidate description; check the full catalog and compatibility at the target revision before installation.
  • This review uses only the supplied static sample. Skills, installer, and CI checks were not run, and external services, costs, and actual outputs were not verified.
Evidence confidence: Low Reviewed Nov 08, 2026 Reviewed revision cbab58188611
See the full review method →
Maintain this project? Add the FARS badge to your README

The badge shows the current review score and verdict and links back to this page. It refreshes about once an hour after a re-review.

FARS

[![FARS review](https://followagents.com/badge/sandbase-skills.svg)](https://followagents.com/en/agents/sandbase-skills)
<a href="https://followagents.com/en/agents/sandbase-skills"><img src="https://followagents.com/badge/sandbase-skills.svg" alt="FARS review"></a>

Recent releases

Latest entries from the project's GitHub Releases. Notes are quoted from upstream as written, not rewritten or tested by FollowAgents.

  1. v0.3.5 SandBase Skills v0.3.5 — verifiable source identity
    Release notes excerpt
    SandBase Skills v0.3.5 — verifiable source identity across Agent Skills hosts This GitHub source release strengthens the portable multi-source-search workflow and makes its runtime requirements explicit. What changed - Canonical source URL identity: the offline evidence-ledger validator now rejects cosmetic duplicates caused by host/scheme case, default ports, fragments, and common tracking parameters. Meaningful query parameters remain distinct. …
    View on GitHub ↗
  2. v0.3.4 SandBase Skills v0.3.4 — native DeepSeek Harness bundle
    Release notes excerpt
    Native DeepSeek Harness bundle SandBase Skills v0.3.4 is directly installable as a DeepSeek Harness plugin bundle. The package declares dsh.bundle and mounts the packaged marketing and research Skill roots through the official @deepseek-ai/dsh-skill-filesystem plugin while retaining DSH's default roots. ```bash dsh plugin --profile web add github:sandbaseai/sandbase-skills ``` The bundle is source-visible at dsh/cordis.patch.yml and covered by a dedicated Node test that verifies both mounted directories. Better discovery …
    View on GitHub ↗
  3. v0.3.3 SandBase Skills v0.3.3 — portable, bounded research
    Release notes excerpt
    What changed SandBase Skills v0.3.3 makes the flagship multi-source research workflow easier to try without an account and more predictable under long searches. - Portable by default: multi-source-search starts with the host agent's own search, page-reading, browser, or academic tools. SandBase is an optional source of additional provider diversity. - Bounded research loops: the Skill now sets a finite search/page budget, stops when further queries are unlikely to add independent evidence, and reports unresolved gaps instead of looping. …
    View on GitHub ↗

FAQ

Does it cost money?
Skills are free and open source (Apache-2.0). multi-source-search runs on host-provided tools with no SandBase account; optional SandBase API calls are usage-based, typically $0.001–$0.01 per call, and a typical research task costs $0.05–$0.20.
What works without a SandBase account?
The flagship multi-source-search Skill is fully usable. Specialized social (e.g. twitter-intelligence), market, and data-provider Skills require SANDBASE_API_KEY to reach those platforms and data sources.
Which agents are supported and where do Skills install?
Any agent implementing the Agent Skills specification: Claude Code (~/.claude/skills/), Codex (~/.codex/skills/), Cursor (~/.cursor/skills/), Gemini CLI (~/.gemini/skills/), plus OpenClaw, Hermes, Amp, and Devin via npx skills add; there are also a Copilot CLI plugin and a native DeepSeek Harness bundle.
Does the evidence-ledger validator prove conclusions are true?
No. It runs offline and checks internal consistency — rejecting unknown/duplicate sources, inflated confidence, unused evidence, and conflicted high-confidence claims — but it does not claim that a source is true.
What is required when adding or changing a Skill?
Update the matching catalog/skills/ entry, the skills.json registry manifest, and the relevant generated packaging, then run npm test, npm run marketplace:check, npm run agent-plugin:check, and python3 scripts/skillpack.py validate; see CONTRIBUTING.md.
Is SandBase Skills Library open source?
Yes. The source code is on GitHub (sandbaseai/sandbase-skills) under the Apache-2.0 licence.
Is SandBase Skills Library safe to use?
FollowAgents' FARS review (Nov 08, 2026) scored SandBase Skills Library 18/29 on trust and safety. None of the eight safety controls the review checks for was missing outright.
Is SandBase Skills Library still maintained?
Development has slowed: the latest commit on GitHub is from 2026-09-26, within the last six months but not the last 30 days. The latest release is v0.3.5 (2026-08-20).
Who makes SandBase Skills Library?
SandBase Skills Library is published on GitHub by the account sandbaseai. FollowAgents has not verified which person or company is behind that account.
What is SandBase Skills Library's official website?
The project's GitHub repository lists https://skills.sh/sandbaseai/sandbase-skills/multi-source-search as its website. That link is set by the repository owner; the source code itself is on GitHub (sandbaseai/sandbase-skills).
Was this page helpful?
View on GitHub ↗ Install ↓

Host ecosystems

Compare agents like this one

The same FARS review applied across the shortlist this agent qualifies for.

Related concepts