{
  "markdown": "# hiloop skills\n\nOpen-source [Agent Skills](https://agentskills.io) that teach AI coding agents to operate\n**[hiloop](https://hiloop.ai)** — isolated agent sandboxes you can snapshot and branch, with\nrun-scoped observability.\n\nThe skills follow the open Agent Skills standard, so they work across harnesses (Claude Code, Cursor,\nCodex, and others). They drive hiloop through the `hiloop` **CLI** — the supported agent interface —\nnot an MCP server: a CLI the agent already knows how to drive costs far less context than loading tool\ndefinitions every turn.\n\n## What's here\n\n| Skill | Use it to |\n|---|---|\n| [`autoresearch`](skills/autoresearch/SKILL.md) | Run an autonomous research loop with evolving idea cards, scored experiments, an ensemble, and a leaderboard |\n| [`authenticating`](skills/authenticating/SKILL.md) | Sign in with `hiloop login` (or a key), verify identity, mint and revoke keys |\n| [`operating-sandboxes`](skills/operating-sandboxes/SKILL.md) | Create, inspect, exec, stop, start, and delete sandboxes; follow conditional references for sessions, snapshots, and devboxes |\n| [`managing-volumes`](skills/managing-volumes/SKILL.md) | Publish and version large data once, mount it into many sandboxes |\n| [`managing-secrets`](skills/managing-secrets/SKILL.md) | Manage write-only credentials; delivery remains fail-closed until the proof-bound gateway ships |\n| [`launching-as-workloads`](skills/launching-as-workloads/SKILL.md) | Launch a run as a registered machine identity (a workload) and control who may launch as it |\n| [`querying-observability`](skills/querying-observability/SKILL.md) | Capture a run, query its telemetry with SQL, and follow it live |\n| [`annotating-runs`](skills/annotating-runs/SKILL.md) | Stamp structured judgments (outcome / score) you can filter and aggregate on |\n| [`reporting-product-bugs`](skills/reporting-product-bugs/SKILL.md) | Report a hiloop bug (or send product feedback) to the hiloop team with `hiloop feedback` |\n\n[`AGENTS.md`](AGENTS.md) is the whole-product orientation an agent reads first.\n\n## Quickstart\n\n**1. Install the CLI** (single static binary):\n\n```sh\ncurl -fsSL https://hiloop.ai/install.sh | sh\nhiloop --version\n```\n\n**2. Point the CLI at your deployment, then authenticate** — the built-in default edge is not live,\nso save a context (or set `HILOOP_API_URL`) first. `hiloop login` is the default (`--device` on a\nremote box with no local browser):\n\n```sh\nhiloop config set-context my-deployment --api-url https://api.example.com\nhiloop config use-context my-deployment\nhiloop login\nhiloop whoami\n```\n\n…or headless (an agent / CI), skip the browser with a key: `export HILOOP_API_KEY=\"hil_…\"`.\n\n**3. Upgrade the CLI, then install the skills.** Choose your harness, or install every supported\ntarget:\n\n```sh\nhiloop skills install claude-code  # cursor | codex | gemini | copilot\nhiloop skills install all\n```\n\n`copilot` installs into the current repository; the other targets install for your user. Full target\npaths, native alternatives, and unsupported-harness fallback instructions are in\n[`SETUP.md`](SETUP.md).\n\nThe CLI pins a verified bundle release.\n\nThen ask your agent to capture a run and query its telemetry — the skills guide the rest.\n\n> **Sandbox status.** Core lifecycle and buffered exec are served. Optional capabilities are\n> deployment-dependent and refuse explicitly when absent; see the capability table in\n> [`operating-sandboxes`](skills/operating-sandboxes/SKILL.md). Every create returns an ambient run,\n> with managed capture for explicit entrypoints, exec, SSH output, cooperative HTTP, and OTLP.\n\nFor autonomous metric optimization, point the agent at your task, fixed dataset/scorer, and the\n[`autoresearch`](skills/autoresearch/SKILL.md) skill. Watch it with the public\n[`fleet-dashboard`](tools/fleet-dashboard/README.md).\n\n## Develop\n\nThese skills mirror the `hiloop` CLI surface and the published guides at https://docs.hiloop.ai. Keep\nskill bodies under ~500 lines and push dense schemas into `references/`, per the Agent Skills best\npractices. When the CLI changes, update the matching skill in the same change.\n\n`scripts/check-skills.sh` is a lightweight drift guard (run in CI on every change and weekly): it\nchecks each skill's frontmatter/length/links and — when a `hiloop` binary is on `PATH` — that every\n`hiloop …` command a skill mentions still exists, derived live from `hiloop --help` (no hardcoded\nlist). Run it locally with `./scripts/check-skills.sh`.\n\n## License\n\nApache-2.0 (see `LICENSE`).\n",
  "bytes": 4522,
  "sha": "b4732665aaa570e38a41de799b87890bf116ec02f79cb1e9b77da1e47ceae433",
  "repo_slug": "hiloopai/skills",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/plg_hiloopai_skills_45ab8a06/readme"
}