Bundles of commands, skills, hooks, and agents you install in one go.
claude plugin install caceras/repo-mapA curated forum and registry for the AI-agent ecosystem. Browse Claude Code plugins, MCP servers, skills, hooks, and full agents the community is actually running — vote them up, read the threads, install with one line.
Scans a monorepo and produces a typed graph of packages, apps, and workspaces the agent can navigate.
plugin install caceras/repo-mapGoogle Search Console MCP server — queries, URL inspection, sitemap submissions, all typed.
mcp add caceras/gsc-bridgeTrigger SF crawls and analyze results from inside Claude Code — one skill, full audit suite.
skill install rankfarm/screaming-frogCategories
Every entry is tagged by exactly one primary type, so filters behave and install commands never collide.
Bundles of commands, skills, hooks, and agents you install in one go.
claude plugin install caceras/repo-mapModel Context Protocol servers that add tools, resources, and prompts.
claude mcp add caceras/gsc-bridgeScoped, trigger-on-context capabilities — composable inside any plugin.
claude skill install rankfarm/screaming-frogShell commands that fire on agent events — format, validate, notify.
claude hook add formatter.on_editFull autonomous agents with their own system prompts and tool allowlists.
claude agent install caceras/code-reviewerSlash commands — quick, named prompts you invoke in chat.
/deploy · /crawl · /briefFeatured this week
Produces a typed graph of your monorepo — packages, apps, workspace boundaries — that the agent can navigate without re-globbing.
plugin install caceras/repo-mapTyped Google Search Console tools — queries, URL inspection, sitemap submit — wired through a single MCP connection.
mcp add caceras/gsc-bridgeReference MCP server for the shadcn/ui registry. Read components, fetch blocks, apply themes — all with installation metadata.
mcp add shadcn-official/shadcnThree slash commands — /commit, /commit-push-pr, /clean-gone — that encode a whole team's git etiquette.
plugin install caceras/commit-commandsRuns, inspects, and self-repairs Playwright e2e suites from inside the agent. Pairs with any Next.js or Astro app.
skill install benchyard/playwright-e2ePostToolUse hook that runs Biome or Prettier on every file the agent touches — and re-runs type checks.
hook add tidyhooks/formatter.on_editTrending threads
How it works
Point agentplatform.forum at a public repo. We sniff the plugin.json / mcp manifest / skill frontmatter and draft the listing for you — you confirm the category, tags, and description.
Each entry gets its own thread. Votes are capped to one per agent-dev account. Comments render the same markdown your agent already uses, so pasted SKILL.md blocks look native.
Every listing shows a copy-ready install command for Claude Code. One line, no DNS games, no wrapper scripts. Listings flagged verified go through a runtime-compat check first.
Install example
plugin install caceras/repo-mapmcp add caceras/gsc-bridgeskill install rankfarm/screaming-frogThe house rules
Every listing carries a typed manifest — category, runtime, scopes, permission mode — so filters stay honest.
Exactly one install command per listing. Mirrors the author's repo, never a wrapper.
Verified badge requires a green run against the current Claude Code runtime + MCP spec.
Threads are co-located with the entry. Bugs, migration notes, and show-and-tell stay with the code.
One upvote per agent-dev account per listing. No bots, no packs, no vote gaming.
Pinned weekly thread where people post what they actually installed and why.
Pricing
Browsing, installing, posting, voting. The whole public forum.
For agent developers who publish — analytics, verified badge, priority review.
For teams shipping a catalog — private mirrors, team seats, compliance.
The forum is free. Listings are free. The only thing we ask is that you show up with a repo that works and a line or two about why you built it.