Block Agent Skills
https://github.com/block/agent-skills- Category
- AI Agents
- Rank
- No. 1560Tools index
Previous survey · No. 1568 ·
- Pricing
- Open Source
- Type
- TOOL
- Builder
- block
- GitHub
- 24 stars
- Date
About
Marketplace for agent skills, a hub for sharing and discovering reusable agent capabilities.
What it does
Block Agent Skills packages focused instructions, checklists, scripts, templates, and examples that compatible AI agents can load for specific workflows. Each skill uses structured metadata plus readable Markdown, while repository automation checks submissions for expected structure and suspicious content.
Why it's ranked here
The project pairs a simple contribution model with unusually concrete safeguards. Submissions face frontmatter, file type, size, symlink, binary-content, remote-execution, prompt-injection, and encoded-payload checks. The supplied catalog is small in this snapshot, but its operating model is clear and inspectable.
What's good
Authors can explain a workflow and bundle supporting text, code, templates, or configuration without adopting a complex framework. Validation rejects risky filenames, executable formats, oversized files, path traversal, and binaries disguised as text. Security scanning also produces JSON and SARIF reports for automation.
Tradeoffs
Installation depends on an external skills command and a compatible agent with its skills extension enabled. The metadata parser accepts only a limited YAML subset. Security detection relies on patterns and entropy thresholds, so warnings require human judgment and novel threats may evade the checks.
How to use it well
Use it when a team wants versioned, reviewable agent playbooks for repeatable engineering or operational work. Read a skill before installation, then keep its scope narrow and include verification criteria. It does not replace an agent runtime, package manager, or independent security review.
Technical notes+
README.md documents installation through npx skills add and compatibility with Goose, Claude Desktop, and other skill-aware agents. scripts/validate-skills.mjs is a Node.js CLI that discovers root skill directories, parses required frontmatter, blocks symlinks and disallowed file types, caps each file at 1,000,000 bytes and each skill at 400 files, and samples content for binary signatures. scripts/skills-security-scan.mjs scans selected text files for remote execution, prompt injection, encoded blobs, entropy, and package lifecycle hooks, with JSON and SARIF output; the supplied excerpt is truncated. rp-why/scripts/goose_skill.py exposes Goose commands and a direct Python CLI, while rp-why/scripts/rp_why_core.py centralizes its classifiers. CONTRIBUTING.md defines the submission structure and review expectations.
Observed
- License
- Apache License 2.0
- Content and implementation languages
- Markdown skills with JavaScript and Python tooling
- Install surface
- Installed through the npx skills add command using a repository URL and skill name
- Agent support
- Goose, Claude Desktop, and other agents that support agent skills
- Interfaces
- Skill documents, Goose slash commands, and command-line scripts
- Repository controls
- Automated structure validation plus security scanning with JSON and SARIF reports
Read from README.md, scripts/validate-skills.mjs, scripts/skills-security-scan.mjs, rp-why/scripts/goose_skill.py, rp-why/scripts/rp_why_core.py, rp-why/scripts/growth_nudge.py, rp-why/scripts/rp_why_baseline.py, LICENSE, CODEOWNERS, GOVERNANCE.md, renovate.json, CONTRIBUTING.md, beads/SKILL.md, rp-why/SKILL.md.
What it can do
Browse and discover agent skills
Search criteria or category filters → List of available agent capabilities with descriptions
Upload and share agent skills
Agent skill code, documentation, and metadata → Published skill listing in marketplace
Download and integrate agent capabilities
Selected skill from marketplace → Reusable agent skill package
Rate and review agent skills
Skill rating and written feedback → Public review and rating on skill listing
Search for specific agent functionalities
Keywords or capability requirements → Filtered results matching search criteria
View skill documentation and usage examples
Selected agent skill → Implementation guides and code examples
Tags
Tech Stack
Comments (0)
No comments yet
Editorially curated, with community endorsements as a secondary signal. Corrections welcome.