Workflow plugins

Claude Code and Codex plugins their authors filed under "workflow". All categories.

Non-interactive delegation to copilot/agy CLI agents, restricted to top-level sessions, with a log folder/filename/content convention for cross-session context.

1 skill

LawrenceCHH/agent-delegation 0

Scaffold a personal, git-excluded local backlog (BACKLOG.local.md) in the current repo, and wire the maintenance convention into CLAUDE.local.md.

1 skill

rinman24/claude-skills 0

Sets up a read-only IAM role and a bootstrap identity for AI agents, capped by a permissions boundary and distinguishable in CloudTrail. Bundles the setup script and the Claude Code skill.

1 skill

iamalisson/aws-readonly-agent 0

How to work GitHub issues, branches and pull requests in this owner's repos: feat/<id>-<name> branches, PR-by-default with Closes #<id> first, the area:* / agent-ready label taxonomy, native issue relationships (sub-issue / blocked-by), and the autonomous worktree-subagent workflow for agent-ready issues.

1 skill

Akirathan/claude-plugins 0

Run one high-context orchestrator that fans out autonomous coding agents by dependency wave and gates every PR through multi-model adversarial review (Phase R / Phase G). Never auto-merges.

1 skill

Blink-Build-Studios/orchestrating-parallel-workstreams 0

How to run large, multi-part, or open-ended work: break it into shippable milestones, ask the user at real forks, verify each milestone for real before calling it done, commit as you go, delegate well-specified chunks to background subagents, and treat existing code and the user's request as having reasonable intent behind them.

1 skill

artman41/claude-skills 0

Five-director adversarial board: Phase 1 independent assessment, Phase 2 cross-examination, Phase 3 final vote, Phase 4 audit. Mandatory file:line evidence, domain-bounded veto, rotating skeptic role, no automatic tiebreaker. Auditor enforces director hard rules and verifies high-impact citations.

1 skill

eliranpv11/dissent-board-skill 0

Interactive PR review, one beat at a time. Reconstructs what a change is for, walks it in causal order while the reviewer steers, and ends the way the target calls for: commits on a branch, one GitHub review, or a recorded report.

1 skill

radkode/underwrite 0

hands-on

claude

Keeps the agent to scaffolding, tests, and review while you write the core business logic by hand.

1 skill

xpyx/skills 0

Shrink bloated Claude Code skill/agent/command files using ABCD four-class extraction: bash/Python logic to scripts/, shared params to manifest-schema.json, design notes to DESIGN.md. Dual-track output: stable (SKILL.md ≤220 lines) or growth (≤80 lines + Phase agents). Three modes: full shrink (>500 lines), proposal (200-500 lines), reject (<200 lines).

1 skill

easyfan/skill-shrinker 0

greybeard

claudecodex

Write clear, correct code with less AI slop: surface assumptions, choose the simplest safe path, fix root causes, use agents only when useful, and verify before claiming done.

1 skill

nassiharel/greybeard 0

gsto

claude

Glenn Stovall's workflow tools for Claude Code: research options before committing to a plan

1 skill

GSto/GStos-claude-skills 0

discover

claude

Methodical 5-pass feature-discovery workflow on Claude Code's built-in Workflow engine: system mapping, bounded research, evidence-rule kill-testing, plan tournament, verified execution. No required dependencies; OMC/superpowers/Codex/Gemini are optional boosters.

1 skill

chopra2007/claude-discover 0

Rule-based PreToolUse hooks that block noisy or unrecoverable Bash habits: static echoes, shelling out past dedicated tools, stacked commands, backgrounding, and deleting with rm instead of trash.

hooks

ariesclark/skills 0

github

claude

Redirect GitHub page fetches (WebFetch, curl, wget) to a local shallow clone or gh, instead of lossy page summaries.

hooks

ariesclark/skills 0

知识花园园丁 plugin:只读审计 Obsidian vault、生成提案入 Notion、轮询已批准提案 apply 到 Evergreen、L1 自动补链/补字段、周报、孤岛审查、随手记捕获。分级信任 L0-L3,所有 Evergreen 写入可追溯。

1 skill

sisibeloved/knowledge-garden-skill 0

Resolves pull request code change requests from review comments with a structured completion protocol.

1 skill

skyfackr/skai_agent_skills 0

Project onboarder: scaffolder script + judgment runbook that takes a project to a consistent standard — brand-new, or an existing directory via adopt mode (stale revivals and never-properly-set-up projects: gap-fill only, history immutable, working tree as found). Private GitHub repo (public only on explicit request), pinned per-repo git identity, project CLAUDE.md, serena onboarding (full six-memory set), and a quantitative verification gate.

1 skill

TheMizeGuy/project-onboard-public 0

Standing operating mode that spends your expensive model's tokens on judgment and pushes execution down to cheaper subagents, which compete for a capped per-project roster with real scores, real elimination, and a channel to push back on bad instructions.

1 skill

bazarkua/rank-and-yank 0

Turn broad work into a safe, measurable goal with explicit launch, selective delegation, and verified completion.

1 skill

jayozer/jay-marketplace 0

Run a multi-agent fan-out in zellij: one tab per work package, a Controller that merges, and a CloneManager that keeps a machine pool alive. Includes the pool-operations and verification-discipline references, plus the spawn/message/status scripts.

1 skill

njreid/skills 0

Revise an existing skill so the agent behaves as intended, without the skill swelling — evidence from real transcripts first, knowledge grows freely, procedure asks

1 skill

kedrzu/skills 0

github

claude

Opens the pull request for the current branch with a Conventional Commits title and an evidence-derived description, watches its checks, resolves failures, and merges on confirmation.

1 skill

leonardo-marziali/agentic-development-kit 0

Get parallel session work onto the remote without races — pre-push probes, one-pusher-per-branch claiming, safe rebase-and-retry instead of force, and landing several sessions' branches in one atomic push.

1 skill

young1ll/done 0

Protocol for dispatching, monitoring, and recovering from subagent failures — pre-flight checks, retry policy, fallback strategy, output contracts, and team swarm coordination

1 skill

koromiko/my_claudecode_marketplace 0

Turns a Markdown feature table — or a raw feature list — into an autonomous dev pipeline: clarify each feature, then develop one branch + one PR per feature until the table is done. Works in English or French.

1 skill

barryaliou980/aliou-skills 0

workflow

claude

Workflow utilities — session retrospective with CLAUDE.md, user instructions, and auto memory suggestions

1 command

koromiko/my_claudecode_marketplace 0

Multi-axis code review before merge — correctness, readability, architecture, security, performance — reported as severity-labelled findings and a verdict. Reviews the working diff by default, or a commit, branch, tag, path, or PR. --effort sets depth; --fix applies the fixes in place and leaves them uncommitted. Any model tier; no dependencies.

1 skill

codxse/sdd 0

worktree

claude

Manage git worktrees using git gtr (create, copy files, navigate)

1 skill

clemux/claude-code-plugins 0

Workflow profile for single-maintainer personal projects: temporary branches instead of PRs, automatic code-review and simplification passes, and merge/cleanup conventions.

hooks

samalone/claude-plugin-marketplace 0