AI use cases Agent skills & MCP Repeatable agent workflows with Agent Skills
Repeatable agent workflows with Agent Skills
An Agent Skill is a folder with a SKILL.md file that an agent loads when a task matches it. Package a recurring task as a skill and the agent follows the same steps every time. Below: the skill collections we track, compared on public data, plus prompts and a workflow.
- Tools compared
- 3
- Open source
- 2of 3
- Reviewed
- 23 Oct
For: Developers
- 1SuperpowersSkills-based software development methodology for coding agents62No dependencies scanned
- 2Agent SkillsAnthropic's example skills, the Agent Skills spec and a skill template53No issues found
- 3wshobson/agentsPlugin marketplace with agents, skills and commands for several coding tools42No issues found
What the score and shapes mean
Score = public maintenance, adoption and docs signals, as a percentile of 52 tools. Nobody here tested these hands-on. Shapes are the separate safety label: No issues found, No dependencies scanned, Caution.
What it does for you
Developers and teams who repeat the same multi-step tasks with an agent.
- Write down a recurring task once and reuse it
- Share a team workflow as a skill collection
- Adopt a ready-made software development methodology for coding agents
- Install community skills and agents from plugin marketplaces
- Keep long instructions out of every prompt
More about this job
A skill is a folder with instructions (and optionally scripts) that an agent loads only when it is relevant. That makes agent behaviour more consistent: your checklist, your conventions, your definition of done.
In short
- Highest score: Superpowers, 62/100 (#13 of 52 tracked tools).
- Most recent push to GitHub: wshobson/agents on .
- Safety label “No issues found”: Agent Skills and wshobson/agents; the rest show “No dependencies scanned”, see the table.
- Permissive open-source licence: Superpowers (MIT) and wshobson/agents (MIT).
Computed from our data of 2 Nov 2026. Scores rank public GitHub signals and are not a hands-on test.
Workflow in 5 steps
- Pick one task you did three or more times this month.
- Write the steps as you would brief a new colleague, including what "done" means.
- Save it as a skill, run it on a real task and note where the agent deviated.
- Tighten the wording and add examples or scripts.
- Review community skills line by line before using them.
Full steps
- Pick one task you did three or more times this month.
- Write the steps as you would brief a new colleague, including what "done" means.
- Save it as a skill, run it on a real task and note where the agent deviated.
- Tighten the wording and add examples or scripts.
- Review community skills line by line before using them.
Prompts to copy
Starting points we wrote, not tested results.
Turn this checklist into an Agent Skill: name, one-sentence description of when to use it, and numbered steps. Keep it under 60 lines.
2 more prompts
Read the skills installed in this project and tell me which ones overlap or contradict each other.
Follow the skill for writing release notes and apply it to the commits since the last tag.
Watch out for
- Text an agent reads from the web, issues or documents can contain hidden instructions (prompt injection).
- Treat AI changes like a pull request from a new colleague: read the diff, run the tests and keep a human sign-off before anything ships.
- Check each tool's licence before commercial use, and check the licence of any code or text the AI reproduces.
- A skill is instructions plus possibly scripts.
- Licence to check before commercial use: Agent Skills (Apache-2.0 + proprietary).
- No dependencies scannedSuperpowersCode checks clean, no lockfile to check
- No issues foundAgent SkillsNo issues found in this week’s scan
- No issues foundwshobson/agentsNo issues found in this week’s scan
Why these warnings?
- Text an agent reads from the web, issues or documents can contain hidden instructions (prompt injection). Do not give an agent that reads untrusted content write access to anything important.
- Treat AI changes like a pull request from a new colleague: read the diff, run the tests and keep a human sign-off before anything ships.
- Check each tool's licence before commercial use, and check the licence of any code or text the AI reproduces.
- A skill is instructions plus possibly scripts. Read them before installing: a malicious skill can tell your agent to do harmful things.
- Superpowers: code and instruction-file checks found nothing, but there is no runtime lockfile or manifest, so dependencies were not checked
How the tools compare
Open the comparison table
All tools we track for this use case, highest score first. Releases count entries in the public GitHub releases feed over 90 days; issue close time is the median over the last closed issues (shown when there are at least 5, the same minimum the score uses).
| Tool | Role here | Score | Safety | Licence | Last push | Releases, 90 days | Median issue close | Stars |
|---|---|---|---|---|---|---|---|---|
| Superpowers Agent skill | Skills-based software development methodology for coding agents | 62 #13 | No dependencies scanned | MIT | 5 | 6 days | 310,769 | |
| Agent Skills Agent skill | Anthropic's example skills, the Agent Skills spec and a skill template | 53 #25 | No issues found | Apache-2.0 + proprietary | 0 | – | 186,457 | |
| wshobson/agents Agent skill | Plugin marketplace with agents, skills and commands for several coding tools | 42 #38 | No issues found | MIT | 0 | 2 days | 42,088 |
Recommended tools
From our ranked data, highest score first. A high score does not mean a clean scan.
- 62SuperpowersAgent skill · MIT · #13 overall
- 53Agent SkillsAgent skill · Apache-2.0 + proprietary · #25 overall
- 42wshobson/agentsAgent skill · MIT · #38 overall
Questions
Which tool scores highest for repeatable agent workflows with Agent Skills?
Superpowers, with 62/100, rank 13 of 52 tracked tools (data of 2 Nov 2026). The score is a relative percentile on maintenance, adoption and docs, not a hands-on test, and it is separate from the safety label.
Which of these tools are open source?
Permissive open-source licence: Superpowers (MIT) and wshobson/agents (MIT). Check the licence before commercial use: Agent Skills (Apache-2.0 + proprietary).
What is an Agent Skill?
A folder with a SKILL.md file. The file starts with YAML frontmatter that needs only a name and a description of what the skill does and when to use it, followed by the instructions. Anthropic describes skills as folders of instructions, scripts and resources that Claude loads dynamically. The Agent Skills standard is documented at agentskills.io. Source: anthropics/skills
How do I install Anthropic's skills in Claude Code?
Run /plugin marketplace add anthropics/skills, then /plugin install document-skills@anthropic-agent-skills or /plugin install example-skills@anthropic-agent-skills. After that you use a skill by mentioning it in your request. Source: anthropics/skills
Are Anthropic's example skills open source?
Partly. The skills README says many of the skills are Apache 2.0, but the document skills (docx, pdf, pptx, xlsx) are source-available, not open source. Source: anthropics/skills
Does Superpowers only work with Claude Code?
No. The superpowers README has install steps for more than a dozen coding tools, including Claude Code, Codex, Cursor, Gemini CLI, GitHub Copilot CLI and OpenCode, and says to install it separately for each one you use (checked 23 Oct 2026). Source: obra/superpowers
What is in wshobson/agents?
One collection of plugins, agents, skills and commands for Claude Code, OpenAI Codex CLI, Cursor, OpenCode, Antigravity CLI, GitHub Copilot and Pi. Its README counts 94 plugins, 202 agents, 185 skills and 105 commands (checked 23 Oct 2026). See agents for score and safety. Source: wshobson/agents
Related jobs
- Give AI agents new abilities with MCP and skills
- Turn recorded work into reusable agent skills
- AI pair programming in the terminal
Sources for tool facts
obra/superpowers · anthropics/skills · wshobson/agents. Data fetched 2 Nov 2026. Scores and labels are automated, relative and not a hands-on test.
Get the board every Monday
One email a week: the new top 10, the biggest score changes and every safety label that changed.
Free. Double opt-in: you confirm by email first, unsubscribe in one click.