{
  "markdown": "# 5dive skills\n\nSkills published by [5dive](https://5dive.ai?utm_source=github&utm_medium=referral&utm_campaign=skills-readme) for use with\n[skills.sh](https://skills.sh/) and any agent harness that loads\n`SKILL.md` bundles (Claude Code, openclaw, hermes, …).\n\n| Skill                  | Purpose                                                       |\n| ---------------------- | ------------------------------------------------------------- |\n| [`5dive-cli`](./5dive-cli) | Drive the local `5dive` CLI on a 5dive runtime VM. Lets an agent spawn, inspect, send to, and tear down sibling agents on the same host. |\n| [`diagnose`](./diagnose) | Self-diagnosis for 5dive agents. Runs auth, service, disk, memory, and skill-integrity checks on the VM and returns a root-cause report — so the agent can fix problems itself instead of asking the user. |\n| [`compile-knowledge`](./compile-knowledge) | Compile durable knowledge into interlinked-markdown stores the \"karpathy method\" way — atomic files, `[[links]]`, a maintained index — so the agent gets smarter over time instead of relearning. Memory-first; shared wiki only for teams. |\n| [`openagent`](./openagent) | Author your own OpenAgent persona and mint your shareable holo trading card. Self-service: write a `<id>.persona.yaml`, validate it, check your rarity tier (Common→Mythical) + completeness, render the PNG card, and PR into the character-packs registry. |\n| [`copywriting`](./copywriting) | Write, rewrite, and improve conversion-focused marketing copy for any page — homepage, landing, pricing, feature, product. Headlines, CTAs, value props, taglines, hero sections. |\n| [`no-ai-slop`](./no-ai-slop) | Edit drafts into sharper, more human writing while preserving the writer's voice, or detect AI-slop patterns without rewriting. Use to make a draft clearer, more direct, and less AI-sounding, or to check whether writing reads as AI. MIT, courtesy of Peter Yang. |\n| [`ad-creative`](./ad-creative) | Generate and iterate ad creative at scale — headlines, descriptions, and primary text for paid platforms — and refine based on real performance data. |\n| [`loops`](./loops) | The full lifecycle for agentic loops — recurring agents packaged as a portable [`LOOP.md`](https://agenticloops.dev?utm_source=github&utm_medium=referral&utm_campaign=skills-readme). Find + install + run an existing loop from the directory, or author a new one when nothing fits. The loop-level analogue of `find-skills` + `skill-creator` in one. |\n| [`playwright-e2e`](./playwright-e2e) | End-to-end testing and click-verification of web apps with Playwright. Install, write specs, drive authenticated pages, screenshot, and run in CI. Use to prove a web change actually works in a real browser before shipping. |\n| [`code-review`](./code-review) | Review a diff or pull request for correctness first, quality second. Logic bugs, edge cases, error handling, races, missing tests, API misuse, performance, readability, with a severity model and clear approve vs request-changes verdict. |\n| [`verify`](./verify) | Grade a delivered claim against the artifact instead of against the report of it. Split a claim into checkable assertions, grade each against a named instrument, and emit a three-state verdict — pass, fail, or not-reached — with what you did *not* check stated explicitly. |\n| [`deep-research`](./deep-research) | Run structured, multi-source research that produces a trustworthy, cited synthesized answer rather than a pile of links. Scope the question, fan out across sources, judge source quality, adversarially verify each key claim, separate fact from inference, and report with explicit confidence. |\n\n## Quick start (any coding agent)\n\nInstall straight from the [skills.sh](https://skills.sh/) directory with the open\n`skills` CLI — works in Claude Code, Codex, Cursor, and any harness that loads\n`SKILL.md`:\n\n```bash\n# mint your own OpenAgent persona + shareable holo card\nnpx skills add 5dive-ai/skills --skill openagent\n\n# find, install, and run agentic loops (recurring agents)\nnpx skills add 5dive-ai/skills --skill loops\n\n# or pull the whole set\nnpx skills add 5dive-ai/skills --skill '*'\n```\n\n`openagent` and `loops` are the best starting points — they work in any agent,\nno 5dive account required. [`loops`](./loops) pairs with the open\n[agenticloops.dev](https://agenticloops.dev?utm_source=github&utm_medium=referral&utm_campaign=skills-readme)\ndirectory + `LOOP.md` spec.\n\n## Install on a 5dive agent\n\nFrom the dashboard: **Agents → Connect skills → 5dive-cli → Install**.\n\nFrom the host:\n\n```bash\nsudo 5dive agent skill <agent-name> add \\\n  --source=5dive-ai/skills \\\n  --skill=5dive-cli\n```\n\nOr via the upstream `skills` CLI from any agent's home:\n\n```bash\nnpx skills add https://github.com/5dive-ai/skills --skill 5dive-cli --agent claude-code --yes\n```\n\n## License\n\nMIT.\n",
  "bytes": 4847,
  "sha": "3d304257a5795e79d45ab1d9612070314073bcda10530d59b0bec04e0de6152a",
  "repo_slug": "5dive-ai/skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/skl_5dive_ai_skills_5dive_cli_c245aa0c/readme"
}