@delta-agentcloud/cli 0.0.0-stage → 1.1.22

package/README.md

@@ -1,3 +1,499 @@
-# Temporary Holding Version
+# delta-cli
 
-This version is a temporary placeholder for this package. An operational version to replace this has been submitted for review and is awaiting a staged release.
+Command-line tool for the Delta Agent Cloud, used to manage Delta Sandbox GPU/CPU compute workloads and invoke Science tools.
+
+## Installation
+
+### Option 1: npx one-click install (recommended)
+
+```bash
+npx @delta-agentcloud/cli@latest install
+```
+
+This command performs, in order:
+1. **Global install** of `@delta-agentcloud/cli` (includes the Go binary)
+2. **Select AI tool platforms** (multi-select): generic agent dir `~/.agents/skills/` (Codex/Cursor/OpenCode), Claude Code `~/.claude/skills/`, OpenCode native `~/.config/opencode/skills/`, Cursor `~/.cursor/skills/`, Memento-S dev mode `~/memento_s/skills/`
+3. **Install AI Skills** into the selected platforms' skills directories
+4. **Initialize config** automatically writing the default server address (no manual base_url entry)
+5. **Authenticate** `delta-cli auth login` (interactive; skipped when not a TTY)
+
+AI agents can also install directly with:
+
+> Install delta-cli for me: https://github.com/delta-agentcloud/delta-cli
+
+### Uninstall
+
+```bash
+npx @delta-agentcloud/cli@latest uninstall
+```
+
+Cleans up the global package, AI Skills from all known platforms, and the config file.
+
+### Upgrade
+
+```bash
+# Upgrade CLI + AI Skills
+delta-cli upgrade
+
+# Only check whether an update is available (do not upgrade)
+delta-cli upgrade --check
+
+# Only upgrade the CLI binary
+delta-cli upgrade --cli-only
+
+# Only upgrade AI Skills
+delta-cli upgrade --skills-only
+```
+
+Upgrading the CLI pulls the latest version via `npm install -g @delta-agentcloud/cli@latest`; upgrading Skills reinstalls AI Skills via `npx skills add`.
+
+### Option 2: Download the binary directly
+
+Download the archive for your platform from [GitHub Releases](https://github.com/delta-agentcloud/delta-cli/releases) (below, `latest` points to the newest version):
+
+```bash
+# macOS (Apple Silicon)
+curl -L -o delta-cli.tar.gz https://github.com/delta-agentcloud/delta-cli/releases/latest/download/delta-cli-darwin-arm64.tar.gz
+tar -xzf delta-cli.tar.gz
+sudo mv delta-cli-darwin-arm64 /usr/local/bin/delta-cli
+rm delta-cli.tar.gz
+
+# Linux (AMD64)
+curl -L -o delta-cli.tar.gz https://github.com/delta-agentcloud/delta-cli/releases/latest/download/delta-cli-linux-amd64.tar.gz
+tar -xzf delta-cli.tar.gz
+chmod +x delta-cli-linux-amd64
+sudo mv delta-cli-linux-amd64 /usr/local/bin/delta-cli
+rm delta-cli.tar.gz
+
+# Windows (AMD64, PowerShell)
+Invoke-WebRequest -Uri https://github.com/delta-agentcloud/delta-cli/releases/latest/download/delta-cli-windows-amd64.zip -OutFile delta-cli.zip
+Expand-Archive -Path delta-cli.zip -DestinationPath . -Force
+Move-Item -Path delta-cli-windows-amd64.exe -Destination delta-cli.exe
+Remove-Item -Path delta-cli.zip
+# Put delta-cli.exe in a directory on your PATH
+```
+
+### Option 3: One-line install script (works without npm)
+
+The install script **auto-detects the environment**:
+- **npm present** → installs the `@delta-agentcloud/cli` npm package from the official npm registry
+- **no npm** → downloads the binary directly (PowerShell + curl on Windows; bash + curl + tar on macOS/Linux, both built in), **no Node.js/npm required**
+
+```bash
+# Linux / macOS
+curl -L https://raw.githubusercontent.com/delta-agentcloud/delta-cli/main/install.sh | bash
+
+# Windows (PowerShell)
+Invoke-RestMethod -Uri https://raw.githubusercontent.com/delta-agentcloud/delta-cli/main/install.ps1 | Invoke-Expression
+```
+
+The script downloads the binary directly from GitHub Releases and verifies the SHA-256. Use the `DELTA_CLI_VERSION` environment variable to pin a version.
+
+### Option 4: npm
+
+```bash
+npm install -g @delta-agentcloud/cli
+```
+
+`postinstall` downloads the binary for the current platform:
+
+1. The mirror specified by the `DELTA_CLI_MIRROR` environment variable (if set; must be HTTPS and allow the full-prefix format, e.g. `https://your-mirror`)
+2. The GitHub source `https://github.com/delta-agentcloud/delta-cli/releases/download/...` (fallback)
+
+If the download fails during install:
+
+- Run `delta-cli` once; it will automatically retry the download.
+- Or set a local archive path before installing:
+
+```bash
+DELTA_CLI_ARCHIVE=/path/to/delta-cli-linux-amd64.tar.gz npm install -g @delta-agentcloud/cli
+```
+
+Common environment variables:
+
+| Variable | Description |
+|----------|-------------|
+| `DELTA_CLI_MIRROR` | Download mirror root, e.g. `https://your-mirror`. The script expands it to the full-prefix format (`<mirror>/https://github.com/...`). Must be HTTPS and allowlisted. |
+| `DELTA_CLI_MIRROR_ALLOWLIST` | Additional allowed hosts, comma-separated. |
+| `DELTA_CLI_ARCHIVE` | Path to a pre-downloaded `.tar.gz` / `.zip` to skip the network download. |
+| `DELTA_CLI_SKIP_POSTINSTALL` | Set to `1` to skip the binary download at install time; `delta-cli` retries at runtime. |
+| `DELTA_CLI_FATAL_ON_ERROR` | Exit non-zero on install failure (default: warn only; useful for debugging). |
+| `DELTA_CLI_DOWNLOAD_TIMEOUT` | Max wait time per download source (ms, default 120000). |
+
+### Option 5: Go Install
+
+```bash
+go install github.com/delta-infra/delta-infra-cli/cmd/delta-cli@latest
+```
+
+## Quick Start
+
+### 1. Initialize configuration
+
+```bash
+delta-cli config init
+```
+
+The server address needs no manual entry: it defaults to `https://delta-infra-nacos.laddery.uk` (an existing config keeps its own base_url); the CLI only prompts interactively for a Bearer Token (optional — you can also configure it later with `auth login`).
+
+To point at a self-hosted server, specify it once via flag or environment variable:
+
+```bash
+# Via flag
+delta-cli config init \
+  --base-url http://your-server/api/v1 \
+  --token your-token
+
+# Via environment variable (API Key or Token, pick one)
+export DELTA_INFRA_BASE_URL=http://your-server/api/v1
+export DELTA_INFRA_API_KEY=your-api-key        # or DELTA_INFRA_TOKEN
+delta-cli config init
+```
+
+The config file is stored at `~/.delta-infra/config.json` (mode 0600).
+
+### 2. Authentication
+
+Two credential types are supported: **API Key** and **Bearer Token** (API Key takes priority).
+
+**Get an API Key**: sign in to the Delta Agent Cloud **web console** and generate a key on the **API Keys** page.
+
+```bash
+# API Key (recommended)
+delta-cli auth login --apikey <your-api-key>
+
+# Token
+delta-cli auth login --token <your-token>
+
+# Interactive login (paste the Key/Token; the CLI detects the format automatically)
+delta-cli auth login
+
+# Show auth status
+delta-cli auth status
+```
+
+You can also provide credentials via environment variables: `DELTA_INFRA_API_KEY=<key>` / `DELTA_INFRA_TOKEN=<token>`.
+
+### 3. Sandbox lifecycle
+
+```bash
+# Create a sandbox with a GPU (--max-life defaults to 30 minutes; raise it for long tasks)
+# ⚠️ Resource semantics: `--memory` is the container CPU memory TOTAL (one per sandbox); `--gpu-mem` is the memory PER CARD (MiB) —
+# `--gpu 2 --gpu-mem 8000` = 8000MiB per card, 16000MiB total (the private OpenSandbox/HAMi backend writes it straight to the HAMi per-card annotation;
+# the cloud RunPod backend treats --gpu-mem as total memory ÷ card count). Per-card memory limit is visible via `sandbox resources` (memory_mib).
+# ⚠️ Resource semantics: `--shm-size` is the container /dev/shm size (default only 64Mi). PyTorch DataLoader(num_workers>0)/
+# multiprocessing, Chromium-based kernels etc. fail with "No space left on device"/"Bus error" when shared memory is insufficient; raise it then.
+# The unit is mandatory (4G/512M are normalized to 4Gi/512Mi; a bare number is rejected by the server) and must not exceed --memory.
+# `--devices` requests host devices (comma-separated, e.g. kvm → /dev/kvm inside the container, for nested virtualization): a privilege escalation,
+# rejected by the server when the allowlist is disabled (never silently ignored). Both apply to private compute only; third-party compute ignores them.
+delta-cli sandbox create \
+  --image image.yangtzeailab.com/opensandbox/pytorch-cuda13:latest \
+  --cpu 4 --memory 16Gi --gpu 1 \
+  --max-life 120
+
+# For DataLoader multiprocessing / nested virtualization: raise /dev/shm and request /dev/kvm (shm must not exceed memory)
+delta-cli sandbox create \
+  --image image.yangtzeailab.com/opensandbox/pytorch-cuda13:latest \
+  --cpu 4 --memory 8Gi --shm-size 4Gi --devices kvm \
+  --max-life 120
+
+# Sandbox that must live long and not be auto-cleaned: add --no-auto-cleanup (only explicit kill/finish destroys it)
+delta-cli sandbox create \
+  --image image.yangtzeailab.com/opensandbox/pytorch-cuda13:latest \
+  --cpu 4 --memory 16Gi --gpu 1 \
+  --no-auto-cleanup
+
+# Run a command (short tasks run synchronously; use run-bg instead of run for long tasks)
+delta-cli sandbox run <sandbox_id> \
+  --command "<command>" \
+  --timeout 3600
+
+# v1.0.85+ sandbox run uses an SSE real-time stream: stdout carries raw data: frames (no trailing envelope); exit_code/log_file are in the complete frame
+# To get the trailing JSON summary in stdout: read the log_file and parse it (sandbox read <id> --path <log_file>)
+delta-cli sandbox run <sandbox_id> --command "echo '{\"status\":\"ok\",\"result\":\"done\"}'" --timeout 60
+
+# Read a result file
+delta-cli sandbox read <sandbox_id> --path /workspace/result.json
+
+# Destroy the sandbox
+delta-cli sandbox kill <sandbox_id>
+```
+
+## Command list
+
+### Global commands
+
+| Command | Description |
+|---------|-------------|
+| `delta-cli config init` | Initialize the config file |
+| `delta-cli config show` | Show current config (sensitive fields masked) |
+| `delta-cli config set <key> <value>` | Set a config value |
+| `delta-cli config remove <key>` | Remove a config value (supports `token`, `apikey`) |
+| `delta-cli auth login` | Interactive login (detects API Key / Token automatically) |
+| `delta-cli auth login --apikey <key>` | API Key authentication (recommended) |
+| `delta-cli auth login --token <token>` | Token authentication |
+| `delta-cli auth status` | Show auth status |
+| `delta-cli upgrade` | Upgrade the CLI and AI Skills to the latest version |
+| `delta-cli upgrade --check` | Check whether an update is available |
+| `delta-cli upgrade --cli-only` | Only upgrade the CLI binary |
+| `delta-cli upgrade --skills-only` | Only upgrade AI Skills |
+| `delta-cli usage` | Show current usage and remaining quota |
+| `delta-cli --version` | Show version |
+
+### Sandbox commands
+
+> **Tip**: `sandbox list` (list sandbox instances) differs from `sandbox ls <id>` (list a directory); don't confuse them.
+
+| Command | Description |
+|---------|-------------|
+| `delta-cli sandbox list` | List the current user's sandbox instances |
+| `delta-cli sandbox providers` | Show available compute backends |
+| `delta-cli sandbox resources [--provider opensandbox\|autodl]` | Show remaining requestable resources (GPU/memory/cores) per backend |
+| `delta-cli sandbox images` | Show the available image list (each entry has `image_name`/`tags`/`description`/`supports_shared_workspace`; `supports_shared_workspace=true` means the compute supports a shared workspace (opensandbox/autodl) and can pair with `--shared-workspace`; the actual seed/write-back is decided at create time by probing for rclone inside the image) |
+| `delta-cli sandbox recommend --cpu N --memory XGi` | Get a resource configuration recommendation |
+| `delta-cli sandbox create --image <img> [--shm-size XGi] [--devices kvm] [--no-auto-cleanup] [--shared-workspace=true\|false]` | Create a sandbox container (`--no-auto-cleanup` disables auto-cleanup; `--shared-workspace` defaults to true: enables the shared workspace (seed/write-back determined at create time by probing for rclone in the image; skipped when absent, data not persistent across sandboxes); false skips seed/write-back, suited to fast tasks; `--shm-size` raises /dev/shm (default 64Mi, **must carry a Gi/Mi unit** and be ≤ `--memory`, common for PyTorch DataLoader/multiprocessing); `--devices` is a comma-separated list of host device tokens (`kvm` → /dev/kvm, nested virtualization), **a privilege escalation** the server **explicitly rejects** when the allowlist is off; `--shm-size`/`--devices` apply to private compute only, third-party compute ignores them) |
+| `delta-cli sandbox extend <id> --max-life N` | **Dynamically extend** a running sandbox's max-life (increase or decrease; N must exceed elapsed minutes and be ≤10080; use `kill`/`finish` for early reclaim); without `--max-life` it queries the current lifecycle (elapsed/remaining/expiry); `status` output includes a `data.lifecycle` summary (elapsed/remaining/expiry; omitted when the lifecycle API is unavailable) |
+| `delta-cli sandbox connect <id>` | Connect to a sandbox |
+| `delta-cli sandbox status <id>` | Show sandbox status |
+| `delta-cli sandbox finish <id>` | Save results and destroy |
+| `delta-cli sandbox kill <id>` | Destroy a sandbox |
+| `delta-cli sandbox run <id> --command "..." [--timeout <sec>] [--summary/--no-summary] [--artifacts]` | Run a command synchronously (raw SSE `data:` frames to stdout, no trailing envelope; `exit_code`/`log_file` in the `complete` frame); `--summary`/`--artifacts` only apply on the legacy-server fallback path |
+| `delta-cli sandbox run-bg <id> --command "..." [--timeout <sec>] [--wait] [--summary/--no-summary] [--artifacts]` | Run a command in the background; without `--wait` it returns `{execution_id, sandbox_id}` immediately; with `--wait` it follows via SSE (raw `data:` frames to stdout, no trailing envelope) |
+| `delta-cli sandbox logs <id> --execution-id <eid> [--tail N --grep <pattern> --context N --max-bytes N]` | Show background logs; returns `stderr_size + stderr_tail` by default (avoids context blow-up); filter with `--tail/--grep` |
+| `delta-cli sandbox cancel <id> --execution-id <eid>` | Interrupt a background command |
+| `delta-cli sandbox read <id> --path <path> [--output <local path>] [--tail N] [--grep <pattern>] [--offset N] [--limit N] [--context N] [--max-bytes N] [--parse-json]` | Read a file in the container; `--output <path>` saves locally; filter with `--tail/--grep`; non-UTF-8 files fall back to base64 automatically |
+| `delta-cli sandbox pull <id> --source <sandbox path> --target <local path> [--recursive] [--pattern <glob>]` | Pull files/dirs from the sandbox to local (mirror of `upload`; flag direction reversed: source=remote, target=local); single file or recursive dir; bidirectional sha1 integrity check on both CLI and server |
+| `delta-cli sandbox write <id> --path <path> --source <local path>` | Write a file |
+| `delta-cli sandbox write-multiple <id> --entry <remote path>=<local path>` | Write multiple files |
+| `delta-cli sandbox ls <id> --path <path>` | List a directory |
+| `delta-cli sandbox stat <id> --path <path>` | File metadata |
+| `delta-cli sandbox mv <id> --entry <src=dest>` | Move/rename |
+| `delta-cli sandbox replace <id> --path <path> --old <text> --new <text>` | Replace content |
+| `delta-cli sandbox chmod <id> --path <path> --mode <octal>` | Change permissions |
+| `delta-cli sandbox rm <id> --path <path>` | Delete a file |
+| `delta-cli sandbox rmdir <id> --path <dir>` | Recursively delete a directory |
+| `delta-cli sandbox mkdir <id> --path <dir>` | Create a directory |
+| `delta-cli sandbox search <id> --path <root> --pattern <glob>` | Search files |
+| `delta-cli sandbox working-directory <id>` | Get the sandbox working directory path |
+| `delta-cli sandbox upload <id> --source <local dir> --target <sandbox path>` | Upload a directory (tar.gz + auto-extract) |
+| `delta-cli sandbox expose <id> [--port N]` | Expose a sandbox service port to the public internet (frp tunnel), default port 8700 |
+| `delta-cli sandbox exposes <id>` | List the sandbox's active public port tunnels |
+| `delta-cli sandbox dispose <id> --port N` | Recycle a public port tunnel (specify the port) |
+
+### Science tools
+
+> The Science Server is derived from `base_url` (`{base_url}/science_tool`); no separate `science_base_url` config is needed.
+> The Sandbox API is likewise derived from `base_url` (`{base_url}/sandbox/api/v1`).
+
+| Command | Description |
+|---------|-------------|
+| `delta-cli science list` | List all enabled science tools |
+| `delta-cli science get <tool_name>` | Show details for a tool |
+| `delta-cli science invoke --tool <tool_name> --endpoint <endpoint_name> [--data '{"key":"value"}'] [--params '{"key":"value"}']` | Invoke a tool endpoint |
+| `delta-cli science endpoints list <tool_name>` | List all endpoints of a tool |
+| `delta-cli science task submit --tool <tool> --operation <operation> --input '<json>' [--wait]` | Submit a durable Science task |
+| `delta-cli science task get/list/cancel/invocations/artifacts` | Query, cancel tasks, and read invocation/artifact metadata |
+| `delta-cli science file invoke --tool <tool> --endpoint <operation> --file <path>` | Invoke a controlled file/multipart operation |
+| `delta-cli science file download --tool <tool> --endpoint <operation> --params '<json>' --output <path>` | Download a raw artifact and compute its SHA-256 |
+
+`tools.name` and `tool_endpoints.name` are the only authoritative names for Science tool calls.
+`science list` and `science endpoints list` show the enabled entries from the server database verbatim;
+`science invoke` passes `--tool` and `--endpoint` verbatim, with no alias mapping or legacy-name compatibility.
+
+Since v1.1.1 the online catalog includes `qe` (Quantum ESPRESSO DFT, 20 endpoints: task
+submit/query/logs/artifact/PP library, etc.) and `materials-design` (durable materials agent
+modeling, 7 endpoints: health/capabilities/job-submit(multipart)/job-status/job-cancel, etc.).
+See the Skill docs `skills/delta-science/references/qe.md` and
+`skills/delta-science/references/materials-design.md`.
+The cloud XRD phase-identification, artifact, CIF simulation and refinement contract is in
+`skills/delta-science/references/xrd.md`.
+
+## Output format
+
+All commands return a unified JSON envelope:
+
+```json
+// success
+{"ok":true,"data":{"id":"sb-xxx","status":"running"}}
+
+// error
+{"ok":false,"error":{"type":"auth","message":"not authenticated","hint":"run 'delta-cli auth login'"}}
+```
+
+Error types and exit codes:
+
+| Type | Exit Code | Scenario |
+|------|-----------|----------|
+| validation | 2 | Argument error |
+| auth | 3 | Not authenticated |
+| permission | 4 | Insufficient permission |
+| not_found | 5 | Resource does not exist |
+| network | 6 | Network error |
+| api | 7 | Server error |
+| internal | 10 | Client internal error |
+
+## AI Agent Skills
+
+delta-cli ships built-in AI Agent operation manuals (Skills) that help assistants such as Claude Code use the CLI correctly:
+
+```bash
+# Install all Skills from the repo
+npx skills add delta-agentcloud/delta-cli -y -g
+
+# Install only the unified Science Skill
+npx skills add delta-agentcloud/delta-cli -s delta-science
+```
+
+Skill files live under `skills/`:
+- `delta-shared/` — shared global rules (auth, config, error handling)
+- `delta-sandbox/` — Sandbox operation guide (lifecycle, command routing)
+- `delta-science/` — the single Science entry point, with the unified Delta CLI wrapper, cross-tool workflows, and
+  on-demand references for PubChem, RDKit, pymatgen, XRD, GSAS-II, LAMMPS, Delta-BO, LDM-BO, SynBO, AntBO
+
+Hosts such as Memento only need to select `delta-science`. The Skill selects the tool and
+operation from a natural-language research goal and loads only the reference for the current
+tool, avoiding the planner overhead of many independent Skills. Upgrades automatically remove
+the old standalone Science Skills.
+
+### Adding a Science tool or operation
+
+The tool catalog is maintained by the Science Server database. To add a tool, configure a
+`tools` record and its `tool_endpoints` records; set `status=false` to disable. The CLI needs
+no static table sync or new commands. To let an AI auto-select a new tool and safely assemble
+requests, also update the tool-selection table in `skills/delta-science/SKILL.md` and the
+corresponding `skills/delta-science/references/<tool>.md`; the references directory holds only
+real tool docs, never meta-docs such as catalogs, routing, service maps or cross-tool
+workflows. The canonical tool name in the database must exactly match the tool table and its
+reference.
+
+## Development
+
+```bash
+# Local build
+make build
+
+# Run tests
+make test
+
+# Lint
+make lint
+
+# Cross-platform build (produces platform binaries in bin/)
+make release
+
+# Install into the system PATH from source (default /usr/local/bin; macOS / Linux)
+make install
+
+# Custom install prefix
+make PREFIX=$HOME/.local install
+
+# Windows source install (pick one)
+# A: install to GOPATH/bin with go install
+go install github.com/delta-infra/delta-infra-cli/cmd/delta-cli@latest
+
+# B: run the PowerShell script from the repo root to build and add to the user PATH
+# .\install-from-source.ps1
+# .\install-from-source.ps1 -InstallDir "$env:USERPROFILE\bin"
+```
+
+### Daily development flow
+
+After modifying `skills/` or building `delta-cli`, sync to the Memento-S project for it to take effect:
+
+```bash
+# 1. Sync the full Skill set via the install wizard
+node scripts/install-wizard.js
+
+# 2. Build the Windows version with Docker (when no local Go toolchain is available)
+docker run --rm -v "$(pwd):/src" -w /src \
+  -e GOPROXY=https://proxy.golang.org,direct \
+  golang:1.23 \
+  sh -c 'go build -buildvcs=false \
+    -ldflags "-X github.com/delta-infra/delta-infra-cli/internal/build.Version=$(git describe --tags --always --dirty 2>/dev/null || echo dev)" \
+    -o bin/delta-cli-windows-amd64.exe ./cmd/delta-cli'
+
+# 3. Place delta-cli into Memento-S's node_modules/.bin (replacing the old version)
+cp bin/delta-cli-windows-amd64.exe \
+   <memento-s>/electron_rebuild/node_modules/.bin/delta-cli
+
+# 4. Restart the Memento-S sidecar to pick up the new binary
+#   Reload in the Electron UI, or restart the sidecar process manually
+```
+
+> **Note**: after replacing `delta-cli`, if a token is already configured in `~/.delta-infra/config.json`,
+> the Sandbox and Science Skills work immediately; otherwise configure it first with `auth login`.
+
+## Build and Release
+
+### Local build
+
+```bash
+# Local build
+make build
+
+# Run tests
+make test
+
+# Lint
+make lint
+
+# Cross-platform build (produces platform binaries in bin/)
+make release
+```
+
+> `make release` only produces platform binaries (e.g. `bin/delta-cli-linux-amd64`); it does not package them. Use `release.sh` for packaging and publishing.
+
+### One-click release (maintainers)
+
+The `release.sh` script at the repo root drives the full release flow:
+
+```bash
+export GH_TOKEN=<your-github-pat>
+export NPM_TOKEN=<your-npm-token>
+./release.sh v1.1.19
+```
+
+Script steps:
+1. Update the `package.json` version.
+2. Cross-compile the 5 platform binaries (prefers Docker + `make release`, falls back to the local Go toolchain when Docker is unavailable).
+3. Produce `.tar.gz` / `.zip` archives and SHA-256 checksums in `bin/`, and update the root `checksums.txt`.
+4. Create the GitHub Release and upload the binary archives and checksums via `.ci/publish-github-release.js`.
+5. Sync the npm scripts, README and all Skills to the public repo `delta-agentcloud/delta-cli`.
+6. Run `npm publish --access public` to publish `@delta-agentcloud/cli`.
+7. Create the local source commit and Git tag; the source GitLab remote is pushed only when `PUSH_SOURCE=1` is explicitly set.
+
+`gh` CLI and Node.js/npm must be installed, and `GH_TOKEN` and `NPM_TOKEN` configured. See `.gitlab-ci.yml` for the CI flow.
+
+## In-sandbox mirror auto-configuration
+
+At sandbox creation, the server **auto-detects the deployment's network region** (mainland China / overseas) and, when the region is mainland China, injects mirror configuration into the container. No manual configuration is needed — it works on entry.
+
+- **Mainland China (domestic)**: the server injects mirror config so `pip install` / `npm install` / `git clone` / HuggingFace downloads use domestic mirrors.
+- **Overseas**: no mirrors are injected; each package manager uses its official source.
+- The server probes connectivity to `pypi.org` / `npmjs.org` to decide the region (cached; falls back to overseas on probe failure, i.e. fail-open).
+
+Four categories of configuration are covered:
+
+| Type | Description | Example |
+|------|-------------|---------|
+| pip index-url | Python package index | `https://pypi.tuna.tsinghua.edu.cn/simple` |
+| npm registry | Node package source | `https://registry.npmmirror.com` |
+| HF_ENDPOINT | HuggingFace download domain | `https://hf-mirror.com` |
+| git insteadOf | GitHub clone proxy | `https://gh-proxy.com/https://github.com` |
+
+The AutoDL compute backend has mirror injection off by default (overseas platform); it takes effect only when operations explicitly enable it.
+
+> **Note the two separate mirror mechanisms**: this in-sandbox mirror config targets the **package managers inside the sandbox** (pip / npm / git / HF). It is unrelated to the CLI's own download mirror `DELTA_CLI_MIRROR` — `DELTA_CLI_MIRROR` is used only to **install the delta-cli binary** and does not affect the sandbox.
+
+## Tech stack
+
+- **Language**: Go 1.23+
+- **CLI framework**: [Cobra](https://github.com/spf13/cobra)
+- **Config**: JSON file (`~/.delta-infra/config.json`, mode 0600)
+- **Output**: JSON envelope (AI-agent friendly, scriptable)
+- **Distribution**: GitHub Release (tar.gz/zip) + npm wrapper
+- **Architecture**: flat static command tree, Factory dependency injection, lazy initialization
+
+## License
+
+MIT

package/checksums.txt

@@ -0,0 +1,5 @@
+4afefbba3350e2ec327ef06f619a545ebe1d073b1b3e114b168bc89c5d2b65c0  delta-cli-darwin-amd64.tar.gz
+527371efd7cb487b5b9f204cf275909d6e91761d581cc22fed29e299e8fe07e5  delta-cli-darwin-arm64.tar.gz
+bc8a485c30d035b3c48a8f95ce703a89d57f33385a39af49e10bb7da169fce5e  delta-cli-linux-amd64.tar.gz
+c4c7452a629dc4134cd0c8a9d6474f73a99e659ddfb3da365557c6a14521e002  delta-cli-linux-arm64.tar.gz
+3228551b42bdefd49304f7e34e12132eb4563111dd4c7c31c940f002e7815741  delta-cli-windows-amd64.zip

package/package.json

@@ -1,6 +1,49 @@
 {
   "name": "@delta-agentcloud/cli",
-  "version": "0.0.0-stage",
-  "stub": true,
-  "description": "Temporary package placeholder for staged publishing"
-}
+  "version": "1.1.22",
+  "description": "Delta Agent Cloud CLI for sandbox compute and science tools.",
+  "bin": {
+    "delta-cli": "scripts/run.js"
+  },
+  "scripts": {
+    "postinstall": "node scripts/install.js"
+  },
+  "files": [
+    "checksums.txt",
+    "scripts/install.js",
+    "scripts/install-wizard.js",
+    "scripts/run.js",
+    "skills/manifest.json",
+    "skills/delta-sandbox/",
+    "skills/delta-shared/",
+    "skills/delta-science/",
+    "skills/delta-protein/"
+  ],
+  "os": [
+    "darwin",
+    "linux",
+    "win32"
+  ],
+  "cpu": [
+    "x64",
+    "arm64"
+  ],
+  "keywords": [
+    "delta",
+    "sandbox",
+    "cli",
+    "gpu",
+    "compute"
+  ],
+  "license": "MIT",
+  "engines": {
+    "node": ">=18.0.0"
+  },
+  "repository": {
+    "type": "git",
+    "url": "git+https://github.com/delta-agentcloud/delta-cli.git"
+  },
+  "dependencies": {
+    "@clack/prompts": "^1.2.0"
+  }
+}