Best Development workflow skills in 2026: Top 26 Picks, Ranked
Skills that change how the agent works: planning, test-driven development, debugging, code review and git hygiene. Below are the 26 development workflow skills we recommend in 2026, ranked by our editor score, GitHub stars, reader upvotes and how often readers click through.
Last updated October 2026 · 26 listings reviewed
1
A skills library that makes coding agents plan, test-first and debug systematically
★ 294k · +2.7k this week · MIT · updated 2026-09-27
9.0/10
What it does Superpowers, by Jesse Vincent and Prime Radiant, is a skills library that gives a coding agent a full development method rather than one-off tricks. When you start building something, the agent first asks what you are trying to achieve and shows the design in short sections for you to approve. Next it writes a detailed plan of… Read more →
Pros
- Encodes a disciplined engineering workflow
- Very popular and actively maintained
- Works beyond Claude Code
Cons
- Opinionated process can slow down tiny tasks
- Lots of skills to learn at first
Pricing: Free · tdd debugging workflow
2
Small, composable engineering skills: grilling, TDD, specs, tickets and architecture review
★ 274k · +4.9k this week · MIT · updated 2026-09-29
8.4/10
What it does Matt Pocock's Skills is the set of Agent Skills the TypeScript educator uses in his own daily engineering work. They are deliberately small and composable. Each one fixes a common way coding agents go wrong: misreading what you want, using too many words, writing code without feedback loops, or letting a codebase turn into a tangle. Frameworks… Read more →
Pros
- Compact, well-reasoned skills grounded in classic engineering practice
- Available in Claude Code's official plugin marketplace
- Works with any skills-compatible agent via npx skills
Cons
- Engineering skills expect a one-time setup per repo and an issue tracker
- Opinionated workflow may not suit teams already on a framework
Pricing: Open source · tdd planning architecture domain-modeling code-review
3
25 lifecycle skills and commands that make agents spec, test, review and ship with discipline
★ 100k · +1.4k this week · MIT · updated 2026-10-02
8.3/10
What it does Agent Skills, created by Addy Osmani with two collaborators, is a pack of engineering workflows for AI coding agents. It aims to stop agents from taking shortcuts such as skipping the spec, the tests or the review. The skills follow the development lifecycle in six phases: define, plan, build, verify, review and ship. Nine slash commands (/spec,… Read more →
Pros
- Covers the whole lifecycle with verification gates and anti-rationalization tables
- Native install paths documented for many agents
- Includes reviewer personas and reusable checklists
Cons
- Opinionated, process-heavy workflows can slow quick prototypes
- Single-skill npx installs omit the shared references folder
Pricing: Open source · spec-driven-development tdd code-review engineering-workflow slash-commands
4
Makes your coding agent write the minimum code a task needs, without cutting safety
★ 151k · +5.3k this week · MIT · updated 2026-09-14
8.0/10
What it does Ponytail is a skill and plugin that makes a coding agent write only the code a task actually needs. Before writing anything, the agent works down a fixed ladder of questions. Does this need to exist at all? Is it already in the codebase? Does the standard library or the platform already do it? Is there an… Read more →
Pros
- Clear decision ladder that avoids needless dependencies and abstractions
- Review and audit commands for over-engineering
- Benchmarked and supported across many agents
Cons
- Hooks need Node on PATH; instruction-only mode elsewhere
- Aggressive minimalism still needs human review for what it leaves out
Pricing: Open source · yagni minimal-code code-review claude-code-plugin agents-md
5
Turns a codebase into an interactive knowledge graph with tours, search and Q&A
★ 85k · +858 this week · MIT · updated 2026-10-02
8.0/10
What it does Understand Anything turns a codebase into an interactive knowledge graph that you can browse, search and ask questions about. It is aimed at the moment you join a project with hundreds of thousands of lines and no map. The /understand command runs a multi-agent pipeline. Tree-sitter extracts the structure first: imports, exports, definitions, call sites and inheritance.… Read more →
Pros
- Tree-sitter structure plus LLM summaries and guided tours
- Shareable JSON graph with a standalone viewer
- Incremental updates and diff impact analysis
Cons
- Initial analysis can use many tokens on large repos
- Source code is sent to your agent's model
Pricing: Open source · codebase-analysis knowledge-graph onboarding tree-sitter visualization
6
Validated, interactive architecture, sequence and data-flow diagrams compiled from typed JSON
★ 76k · +4.5k this week · MIT · updated 2026-10-02
7.9/10
What it does Archify is an Agent Skill that turns a codebase or a plain-text system description into an interactive diagram. The agent does not draw the SVG itself. It writes a typed JSON description of the system, and Archify's Node.js tooling checks it and compiles it into a single self-contained HTML file. You can ask for a diagram without… Read more →
Pros
- Deterministic compile and validation with machine-readable repair hints
- Single-file HTML output with tracing, themes and PNG/SVG/WebM export
- Architecture Delta view for reviewing design changes
Cons
- Needs Node.js; optional update check pings a manifest unless disabled
Pricing: Open source · architecture-diagram diagrams documentation code-visualization html
7
Terse-output skill plus local proxy that trims the tokens coding agents write and read
★ 109k · +1.1k this week · Apache-2.0 · updated 2026-10-01
7.7/10
What it does Caveman cuts the tokens a coding agent spends, and it comes in two parts. The skill makes the agent answer in clipped, "caveman" prose. Code, commands, file paths and error messages stay exact, and security warnings and confirmations come back in full sentences. The optional proxy runs locally between your agent and the model provider. It compresses… Read more →
Pros
- Keeps code, paths and errors exact while cutting prose
- Local proxy compresses logs, diffs and JSON with recoverable originals
- Honest benchmarks, including where it loses
Cons
- Proxy engine is BSL-1.1 and the CLI sends anonymous telemetry by default
- Skill-only savings in agentic coding are modest
Pricing: Free · token-optimization cost context-compression proxy productivity
8
Turns docs sites, repos, PDFs and videos into Agent Skills and RAG-ready knowledge
★ 15k · +64 this week · MIT · updated 2026-09-30
7.7/10
What it does Skill Seekers is a Python command-line tool that turns documentation and code into Agent Skills and other AI-ready knowledge. You point it at a documentation site, a GitHub repository, a local codebase, a PDF or one of about fifteen other source types. It scrapes and organises the content, can use an LLM or your local coding agent… Read more →
Pros
- About 18 source types and 22 export targets
- Installs generated skills into many agents
- MCP server and quality scoring included
Cons
- Output quality depends on sources and optional LLM enhancement
- Many optional extras make setup sprawling
Pricing: Open source · skill-generation scraping documentation rag mcp
9
Saves AI coding sessions as local Markdown; Lore skill mines them into reusable skills
★ 1.3k · +6 this week · Apache-2.0 · updated 2026-10-02
7.6/10
What it does SpecStory saves the conversations you have with AI coding assistants as Markdown files inside your project, in a .specstory/history/ folder. That gives you a searchable local record of prompts, answers and decisions that would otherwise disappear when a session ends. The getspecstory repository is the public home for the project: it hosts the open-source SpecStory CLI, the… Read more →
Pros
- Local-first capture of sessions from many terminal agents and IDEs
- Lore turns real history into evidence-backed SKILL.md skills
- Open-source CLI with optional cloud search and team sharing
Cons
- Cursor and VS Code extensions are closed source; cloud features need an account
- Saved histories can contain secrets if committed or synced carelessly
Pricing: Freemium · chat-history session-capture skill-generation cli local-first
10
Course-style reference to Claude Code features, workflows and 80+ sourced tips
★ 67k · +652 this week · MIT · updated 2026-10-02
7.5/10
What it does Claude Code Best Practice, by Shayan Raisshan, is a reference and teaching repository for Claude Code. It gathers the product's features, links to the official documentation, and collects tips from the Claude Code team and the community. It also shows each concept as a working example inside the repository. The author says to read it "as a… Read more →
Pros
- Maps every Claude Code feature to docs and a working in-repo example
- Tracks new releases almost daily
Cons
- A learning resource rather than an installable skill pack; some tips come from social posts
Pricing: Open source · claude-code best-practices subagents hooks learning
11
Agent rules and skills distilled from Clean Code, Refactoring, DDD and other classic books
★ 2.9k · +32 this week · MIT · updated 2026-09-10
7.3/10
What it does Agent Rules Books, by Maciej Ciemborowicz, turns the ideas of well-known software engineering books into rule sets that a coding agent can follow. Instead of telling an agent to "follow Clean Code", you give it a concrete list of practices drawn from that book. The rules are the author's own wording, not quotes from the books, and… Read more →
Pros
- Fourteen rule sets in mini, nano and full sizes to fit any context budget
- Plain Markdown with a SKILL.md per book, no scripts to run
Cons
- Rules are one author's interpretation, not official book material; evidence of benefit is a single informal experiment
Pricing: Open source · agents-md refactoring clean-code domain-driven-design architecture
12
Persistent, compressed session memory for Claude Code via hooks, a local worker and MCP search
★ 95k · +495 this week · Apache-2.0 · updated 2026-10-02
7.2/10
What it does Claude-Mem, by Alex Newman, gives Claude Code a long-term memory that lasts across sessions. It hooks into the session lifecycle, records what the agent does with its tools, and uses a model to compress those records into short observations and summaries. It then injects the relevant parts into later sessions. The result is that a new session… Read more →
Pros
- Automatic capture and injection of past session context
- Token-efficient layered search over SQLite and Chroma
- Private tags and local-only mode available
Cons
- Installer defaults to a hosted sign-in trial unless you opt out
- Stores detailed session records; README promotes a crypto token
Pricing: Freemium · memory context-persistence hooks mcp sqlite
13
Self-correcting memory, research wikis and quality-gate hooks for Claude Code
★ 2.9k · +19 this week · MIT · updated 2026-09-29
7.1/10
What it does Pro Workflow is a Claude Code plugin by Rohit Ghumare that gives your sessions a memory. It puts one local SQLite store underneath every session. When you correct Claude, the correction can be saved as a rule and loaded again at the start of every later session, so the same mistake is less likely to come back.… Read more →
Pros
- Corrections persist as rules across sessions
- Full-text research wikis auto-injected when relevant
- Cost, context and permission tooling in one bundle
Cons
- Heavy: dozens of hooks and a local database, may need a build step
- No LICENSE file in the repo despite MIT in package.json; single maintainer
Pricing: Open source · memory hooks quality-gates knowledge-base sqlite
14
JetBrains guidelines that teach AI coding agents to write modern, idiomatic Go
★ 3.8k · +44 this week · Apache-2.0 · updated 2026-09-10
7.0/10
What it does Go Modern Guidelines is an official JetBrains skill/rules package that teaches AI coding agents to write modern, idiomatic Go instead of outdated patterns. Rather than being a standalone application, it's reference material an agent loads and applies automatically when it detects a relevant Go task — for example, favoring the builtin max(a, b) over a manual conditional,… Read more →
Pros
- Officially published and maintained by JetBrains, a credible, established source
- Low-risk: guidance-only, doesn't execute code or modify projects itself
- Documented installers for multiple agent platforms (Junie, Claude Code, Cursor, generic skills)
Cons
- Narrow scope: only useful on Go codebases
- Requires Go 1.25+ on PATH, so it won't help teams on older Go toolchains
Pricing: Free · golang jetbrains coding-standards claude-code cursor
15
Agent Skill that inserts a validated domain model between requirements and code
★ 2.8k · +1 this week · MIT · updated 2026-09-27
7.0/10
What it does TeaQL Agent Kit is a reference "model-mediated harness" for coding agents. Instead of letting an agent jump straight from a plain-language requirement to code, it inserts an inspectable domain model (written in a compact language called KSML) between intent and implementation. A deterministic evaluation service checks the model and returns repair guidance, then a generation service turns… Read more →
Pros
- Real cross-language runtime evidence (7 languages, multiple databases) with a published, dated verification snapshot rather than just claimed support
- Actively maintained (196 commits, MIT licensed, most recent merge September 2026) with clear docs on the model-first workflow
Cons
- Adds real conceptual overhead (a KSML domain-modeling layer) on top of a normal prompt-to-code loop
- Full functionality depends on a hosted, closed generation service (api.teaql.io) rather than being fully self-contained
Pricing: Open source · agent-skill domain-modeling code-generation multi-language typed-contracts
16
MEX Open source
Team memory for engineers and their agents, kept in the repo and shared through Git
★ 1.7k · +17 this week · MIT · updated 2026-10-02
7.0/10
What it does MEX is shared project memory that lives inside your repository and travels through Git. It keeps a team's architecture notes, decisions, requirements and handoffs as readable Markdown next to the code, linked to the actual source through a code graph. Engineers and their coding agents read that context, propose additions or corrections for review, and pass work… Read more →
Pros
- Durable project context and handoffs, no hosted service
- Agents propose knowledge changes for human review
- Works solo and across teammates via Git
Cons
- Pseudonymous telemetry on by default (can be disabled)
- Needs Node.js 22.5+ and a Git repository
Pricing: Open source · memory context-management handoffs code-graph git
17
Turns your coding agent's technical output into plain-language reports and next steps
★ 1.1k · +129 this week · MIT · updated 2026-09-30
7.0/10
What it does Open Steps is a pack of Agent Skills that sits between your coding agent and the humans who need to understand what it did. Instead of scrollback full of diffs, tool calls and jargon, it produces short, plain-language reports: a one-screen "done or not" verdict, numbered steps a non-technical person can follow, and a premortem of what… Read more →
Pros
- Makes agent progress understandable to non-technical stakeholders with an honest verdict, not marketing spin
- MIT licensed and actively maintained, with roughly 950 GitHub stars and recent commits
- Ships as a proper Claude Code plugin with a routing hook, not just loose skill files
Cons
- Full automation (plugin + routing hook) only works on Claude Code; Codex, Cursor and Gemini CLI get the bare skill files copied in manually
Pricing: Free · agent-skill claude-code reporting non-technical workflow
18
Desktop app and CLI that keeps one skills library and symlinks it into 30+ AI coding tools
★ 1.0k · +12 this week · MIT · updated 2026-09-10
7.0/10
What it does Skills Manager is a free desktop app for keeping one library of Agent Skills and sharing it with all the coding tools on your machine. Without it, you end up with separate copies of the same skill in each tool's own folder, such as ~/.claude/skills or ~/.codex/skills, and they drift apart. Skills Manager keeps each skill in… Read more →
Pros
- One source of truth for skills across many tools
- Per-tool enable/disable without deleting files
- Headless skm CLI for scripts and servers
Cons
- macOS build is not notarised; the cask strips quarantine
- Small project; marketplace skills are unvetted community content
Pricing: Free · skills-management desktop-app symlinks cli tauri
19
Output skill that makes coding agents lead with the next action, not filler
★ 53k · +1.7k this week · MIT · updated 2026-09-19
6.6/10
What it does i-have-adhd is a single skill that changes how your coding agent writes its answers. It stops the agent from burying the useful part under context, recaps and polite filler. Once invoked with /i-have-adhd, it stays active for the rest of the session until you say "stop adhd mode" or "normal mode". The first line of each reply… Read more →
Pros
- Clear, testable rules with sensible exceptions
- Install instructions for many hosts
- Plain text, no scripts
Cons
- Single small skill; terse format can drop reasoning you wanted
Pricing: Open source · output-style productivity adhd formatting agent-skills
20
Verification-gated improvement loops as Agent Skills for ML, code, SQL, prompts and research
★ 174 · +2 this week · MIT · updated 2026-06-30
6.6/10
What it does Agent Loop Skills packages iterative improvement loops as open-standard Agent Skills. Each loop is generic: when you invoke it you bind it to your own task by naming the artifact to improve, the command that measures it, and a budget. The agent then proposes one change, runs it in your environment, scores the result against a real… Read more →
Pros
- Every loop keeps a change only when a real metric or test says it improved
- Append-only run ledger records every attempt
- Installs through the standard skills installer or a Claude Code plugin
Cons
- Labelled experimental; parallel subagent roles only work on Claude Code
- Many loops target ML and academic research rather than everyday app code
Pricing: Open source · agentic-loops autoresearch optimization agent-skills red-teaming
21
Claude Code skills and commands for planning, handoffs and building your own extensions
★ 2.0k · MIT · updated 2026-04-01
6.5/10
What it does TÂCHES Claude Code Resources is a personal but reusable pack of slash commands, skills and subagents for Claude Code, packaged as a plugin. Its focus is on meta-work: generating better prompts, planning projects in stages, handing context over between sessions, and building or auditing your own Claude Code extensions. What is inside The skills cover hierarchical project… Read more →
Pros
- Generators and auditors for skills, commands, subagents, hooks and MCP servers
- Structured planning with automatic context handoff
Cons
- One author's workflow; some commands are general thinking prompts
- No updates since April 2026
Pricing: Open source · planning meta-prompting skill-authoring handoff claude-code
22
Evidence-first project planning, course-correction and pre-ship review skill
★ 317 · +3 this week · MIT · updated 2026-08-30
6.5/10
What it does Advise Project Approach is a single Agent Skill for project planning, course correction and pre-ship review. Instead of answering "what stack should I use?" from general impressions, it makes the agent gather your real constraints, find comparable projects, weigh trade-offs, cost and lock-in, and say under which conditions its recommendation would become wrong. It works in three… Read more →
Pros
- Requires comparables, dates and sourced pricing instead of vibes
- Works in pre-build, mid-build and post-build modes
- Documented evaluations and changelog
Cons
- Quality depends on the host agent's web research
Pricing: Open source · planning architecture code-review stack-selection agent-skills
23
Install, manage and share Agent Skills across every major coding agent from one place
★ 260 · +-1 this week · Apache-2.0 · updated 2026-09-19
6.5/10
What it does Skill Flow installs, manages and shares Agent Skills across many coding agents from one place. Instead of copying skills folder by folder into each agent, you import a source once, keep the whole group together, pick the skills you want, and deploy them to several agents at the same time. Sources can be local folders, GitHub repositories… Read more →
Pros
- Deploys one skill set to many agents at once
- Explicit state with manifest, lock and repair tools
- CLI plus a macOS desktop app
Cons
- Does not vet the skills it deploys
- Desktop app is ad-hoc signed, needing a Gatekeeper workaround
Pricing: Open source · skills management multi-agent cli desktop-app
24
Spec-freeze, build, evidence and fresh-verify loop with repo-local task proof
★ 731 · +-1 this week · Apache-2.0 · updated 2026-04-01
6.2/10
What it does Repo Task Proof Loop is a workflow skill for non-trivial coding tasks that keeps the proof of work inside the repository. When you start a task it creates a folder under .agent/tasks/<TASK_ID/, installs project-scoped subagents for Claude Code and Codex, and adds managed blocks to AGENTS.md and CLAUDE.md. The agent then follows a fixed loop: freeze a… Read more →
Pros
- Keeps spec, evidence and verdict inside the repo for audit
- Separates builder and fresh-context verifier
- Generates matching subagents for Claude Code and Codex
Cons
- No installer; copy the folder by hand, and updates have slowed since spring 2026
Pricing: Open source · spec-driven verification subagents evidence workflow
25
Practical Claude Code guide with copyable commands, subagents, hooks and a skill
★ 3.1k · +37 this week · MIT · updated 2026-07-28
6.1/10
What it does Claude Code: Everything You Need to Know, by Wesam Mustafa, is a practical guide to Claude Code paired with a working example project. It walks from installation and a first prompt through prompt workflows, effort levels, slash commands, skills, hooks, subagents, agent teams, dynamic workflows and MCP. Unlike a link list, it explains when to use each… Read more →
Pros
- Explains when to use each extension point
- Runnable example commands, subagents, hooks and workflow
Cons
- A guide with samples, not an installable or versioned package
- Example hook settings need path edits before use
Pricing: Free · claude-code guide hooks subagents slash-commands
26
A Claude skill that writes sharp, tool-specific prompts, including for coding agents
★ 14k · +313 this week · MIT · updated 2026-08-24
6.0/10
What it does Prompt Master is a Claude skill that writes and refines prompts for other AI tools, including the coding agents you use to build software. You describe what you want and the skill produces one clean, copyable prompt aimed at the right target. It runs a structured pipeline on each request: it detects which tool the prompt is… Read more →
Pros
- Structured pipeline: detects the target, asks focused questions, trims tokens
- Profiles for Claude Code, Cursor, Copilot, Codex and other coding tools
- Ships more than a dozen prompt frameworks selected automatically
Cons
- Broad scope covers many non-coding targets you may not need
- Some headline claims are marketing shorthand, not measured figures
Pricing: Open source · prompt-engineering claude-skill templates coding-agents clarifying-questions