Agent Plugins Marketplace

Recently added

Plugins grouped by the day they were added to the directory, newest first.

Sep 17, 2026

  1. agentspec246stars

    Spec-Driven Development framework for Data Engineering — 58 agents, 24 KB domains, 5-phase SDD workflow, 31 commands

    Claude Code16 skills
  2. soujo0stars

    Resumable spec → plan → layer workflow shared by Claude Code and Codex

    CodexClaude Code8 skills
  3. aeon0stars

    Operator console for an Aeon agent instance: enable/schedule/edit skills, wire secrets and channels, debug runs, and mine past Codex/Claude Code chats into scheduled skills.

    CodexClaude CodeAgent Plugins1 skill
  4. port1stars

    A GitHub-driven agent pipeline — plan, implement, review, revise — installable into any repository. GitHub labels are the state machine; a cockpit session dispatches stage agents into isolated worktrees and runs the human gates.

    Claude Code1 skill
  5. mattstack0stars

    The mattstack skills pack: public doors subagent-review-loop, editing-skills, creating-a-pack, extending-a-pack, getting-current-time, wrap-up, the pack's own compiled shepherdr verb (domain-unbound), and the three writing-style presets (sparse, conversational, structured), plus the compile-native engines (work orchestrator, pipeline stages, review verbs, forge adapters, shepherdr) and attachments (model tiering, execution strategies, account pool, review include bodies, review-posting, parameterized-skill primitive) that rt skills compile inlines into team verbs at compile time -- reached by name, not by slash menu.

    Claude Code5 skills1 MCP server
  6. mynd2stars

    Mynd: persistent memory for AI coding agents. Gives Claude Code access to a local memory store that survives across sessions.

    Claude Code1 MCP server
  7. Routes decisions to TypeSafe's Jev model: ranks installed skills for each prompt, and answers the agent's own this-or-that questions when confident.

    Claude Code
  8. taskflow10stars

    PRD/Spec/Plan task workflow skill with host/harness hooks (Claude Code, Codex)

    CodexClaude Code1 skill
  9. Radio interferometric Measurement Set inspector — MCP tools + skills for VLA/MeerKAT/uGMRT

    Claude Code3 MCP servers
  10. Operate Paseo agents and daemon: orchestrate staged workflows, configure agent profiles, and report results.

    CodexClaude Code4 skills
  11. Create new skills, improve existing skills, and measure skill performance. Use when users want to create a skill from scratch, update or optimize an existing skill, run evals to test a skill, or benchmark skill performance with variance analysis.

    Codex1 skill
  12. trantor1stars

    Trantor — the hub-world for AI agent crews: live message bus, presence, project Kanban/flow board + crew orchestration for independent AI coding agents (Claude, Codex, Gemini, Kimi, DeepSeek)

    Claude Code4 skills1 MCP server
  13. Literature, Zotero, PDF, citation, materials, and scientific image tools

    CodexClaude CodeAgent Plugins8 skills1 MCP server
  14. k-mem0stars

    Memory and governance for Claude Code: a read-before-write gate, a context resolver with receipts, tiered project notes, handoffs between sessions, and docs checks that fail when the record and the tree disagree.

    Claude Code4 skills
  15. Research Loop projects, pages and AI work; local methodology figures and consistent research plots.

    CodexClaude Code3 skills1 MCP server
  16. Fully local visual plans and recaps with diagrams, rich diffs, live reload, and a browser feedback loop.

    CodexClaude Code2 skills
  17. Use premium models for judgment while delegating bounded, token-heavy work to cheaper subagents.

    CodexClaude Code1 skill
  18. Budget loop for long-running or parallel Claude work under five-hour and weekly usage limits.

    CodexClaude Code1 skill
  19. recap8stars

    End-of-session recap that reports the result, not the journey, in plain language for a reader who never saw the code.

    CodexClaude Code1 skill
  20. Docs-first discipline for external APIs, packages, providers, and fast-moving product behavior.

    CodexClaude Code1 skill
  21. Build one clean coding plan: explore in parallel, review the draft against the ask, batch the questions, write tickets and resolve spikes, log every decision, then hand off to implementation.

    CodexClaude Code1 skill
  22. Implement an agreed plan hands-free: worktrees and cheap executors under TDD with docs updated in the same diff, an item-by-item check against the plan, an adversarial review, and decisions kept where they belong.

    CodexClaude Code1 skill
  23. Opinionated Go writing discipline covering idiomatic code, errors, tests, comments, and behavior-preserving simplification.

    CodexClaude Code1 skill
  24. Plain-language communication for readers who cannot see the code, using Simplified Technical English and outcome-first explanations.

    CodexClaude Code1 skill
  25. Language-agnostic code simplification that reduces complexity while preserving exact behavior.

    CodexClaude Code1 skill
  26. Re-asks an agent's open questions one at a time, in one fixed shape, with every recommended fix checked first.

    CodexClaude Code1 skill
  27. Comment discipline for code the agent writes: every comment adds information the code does not and describes the current state only.

    CodexClaude Code1 skill
  28. Hostile, bias-free review using independent reviewers, false-positive verification, and validated fixes.

    CodexClaude Code2 skills
  29. kgai0stars

    Shared decision memory for AI dev teams. Your agent records the decisions behind your code — what changed, why, even the dead ends you ruled out — recalls them before changing code, and syncs them across the team without merge conflicts. Local-first, immutable, measured to 1M decisions.

    Claude Code
  30. Your team's coding standards, in context. Connects to a dev-playbook server (run it locally with Docker) over MCP to serve guardrails, definition of done, per-language rules and task workflows - and scaffolds them for a repo that has none. Run /dev-playbook-init to set it up.

    Claude Code2 skills
  31. triforce0stars

    Runs a change through an orchestrated multi-agent workflow — zelda plans and merges on Opus, link executes in isolation on Sonnet, and ganondorf renders a verdict on the merged diff against a frozen, user-sourced criteria list. The audit is required but never blocking, bounded by a monotone per-branch invocation ledger rather than by the reviewer's own judgement, so it terminates instead of finding something every round.

    Claude Code1 skill
  32. kollate0stars

    Capture your Codex conversations into your organisation's Kollate workspace

    CodexClaude Code
  33. clooks0stars

    Set up Clooks and create TypeScript hooks for your agents.

    CodexClaude Code
  34. Project-configured clooks hooks: enforce project scripts, protected paths, and JS/TS package manager choice

    Claude Code
  35. Educational hook pack: lifecycle, config, events, and debugging examples for clooks

    Claude Code
  36. Curated zero-config production hooks for clooks: command safety, git protection, tool hygiene, and tmux notifications

    Claude Code
  37. Aplica y refuerza el uso estricto de español neutro latinoamericano estándar (tuteo profesional, sin voseo ni modismos regionales) en interacciones, documentación, commits y código en Claude Code.

    Claude Code2 skills
  38. Aplica y refuerza el uso estricto de español neutro latinoamericano estándar (tuteo profesional, sin voseo ni modismos regionales) en interacciones, documentación, commits y código en Claude Code.

    Claude Code2 skills
  39. loupe1stars

    Hand pull request review findings to a human with loupe: capture the pull request, file the findings into a local draft, open review for the human, and answer their notes. Never publishes.

    CodexClaude Code1 skill
  40. Create and retrieve from a Foundry IQ knowledge base using File, Azure Blob, or ADLS Gen2, then connect an existing Prompt or Hosted Agent.

    Claude Code1 skill
  41. Spend fewer tokens on Rust: working rust-analyzer LSP, quiet cargo, outline/crate-src/diag tools, and a doctor that checks it all

    Claude Code2 skills
  42. qa-agents0stars

    Multi-agent pipeline for Playwright/POM UI test automation: requirement -> test case -> spec, plus heal/compliance/quality loops. Run /qa-agents:init once per target project before using the other commands.

    Claude Code
  43. Auto Office v3 adaptive orchestration runtime: durable role packets, harness/model/effort routing, adapter trust, quota-aware cost policy, verification, private telemetry, replay, maturity, and isolated self-improvement. Human merge-to-main remains mandatory.

    Claude Code18 skills
  44. Figma → Astro. Turns a Figma frame into Astro code that lands in the target repository's own conventions — deriving those conventions at generation time rather than carrying them, and keeping a memory of what it learned. Built for Iron Software's iron-websites, portable to any Astro repository with a content layer.

    Claude Code19 skills1 MCP server
  45. Daily D-033 BX Self-Service Tool data-pipeline sanity check (DMABGS-3271): verify the two Snowflake tasks succeeded and the output table's max date is fresh with non-zero revenue per brand, then post a pass/fail summary to Slack.

    Claude Code1 skill
  46. Daily D-009 Site Performance data-pipeline sanity check: verify Snowflake tasks succeeded and output table max dates are fresh, then post a pass/fail summary to Slack.

    Claude Code1 skill
  47. Daily D-001 Performance Cube data-pipeline sanity check (Jira DMABGS-3270): verify Snowflake tasks succeeded, each cube slice is fresh with non-zero revenue and spend, and Google/Bing per-source spend is non-zero, reconcile revenue and spend vs. source day-by-day over a rolling ~2 months, then post a pass/fail summary to Slack.

    Claude Code1 skill
  48. Daily D-000 Channel Dashboard data-pipeline sanity check (Jira DMABGS-3269): verify the Snowflake task succeeded and dashboard slices are fresh with non-zero revenue and spend, gate on hourly completeness (all 24 UTC hours of PPC clicks and sessions present) to catch feed interruptions / site downtime / partial-day loads, reconcile revenue and spend vs. source day-by-day over a rolling ~2 months, run a shadow-mode trend/anomaly check on the new data (per channel, same-weekday baseline, US-holiday aware), then post a summary to Slack.

    Claude Code
  49. topo0stars

    Run your Topo outbound workspace from chat: triage replies, clear tasks, review waiting leads, start outreach, source leads, and brief accounts.

    CodexClaude CodeAgent Plugins13 skills1 MCP server
  50. Doctrine patrimoniale et simulateurs fiscaux auditables pour conseillers en gestion de patrimoine. L'IA ne calcule pas : elle pilote des moteurs de calcul vérifiés et cite ses sources.

    Claude Code1 MCP server