Rendu depuis le dépôt source en conservant titres, exemples, code, tableaux, liens et images.
Huawei Cloud Agent Integration (OpenViking Long-Term Memory)
Overview
Integrate and unbind OpenViking long-term memory with coding agents. Agents run in bwrap sandboxes under /root/job-envs/sandboxes/ and use their native mechanism — MCP (mcp__openviking__* tools) or HTTP memory provider — so integration survives agent upgrades.
Integration writes are template-level persistent: config is injected into the agent's start.sh / config templates under /root/template/<agent>/, so sandbox stop + start preserves the integration.
Supported Agents
| Agent | Mechanism | Persistence |
|---|---|---|
| CodeArts CLI | @openviking/opencode-plugin → .codeartsdoer/node_modules/ | Template + live |
| OpenCode | @openviking/opencode-plugin (npm mirror → GitHub fallback) | Template start.sh |
| OpenClaw | clawhub:@openviking/openclaw-plugin + contextEngine slot | Template start.sh |
↳ Optimization: disables session-memory hook + denies memory_search tool (OpenViking handles recall; local memory degrades quality) | ||
| Hermes | Built-in memory.provider: openviking (HTTP REST, no MCP) | Template + live |
| WorkSwarm | Dual-channel: native provider + MCP (13 tools) + code-mode patch | Template + live + runtime patch |
| KimiCode | MCP via mcp.json | Template + live |
| DeepSeek Harness | @openviking/dsh-memory-plugin bundle (on-demand from GitHub) | Template start.sh |
| Prime Agent | @openviking/pi-coding-agent-extension (on-demand from GitHub) | Template start.sh |
Per-agent config details: references/agent-configs.md.
Prerequisites
- OpenViking server running at
http://127.0.0.1:1933(curl -s http://127.0.0.1:1933/health). - Agent sandboxes exist under
/root/job-envs/sandboxes/(managed by job-env-manager). - Host tools:
curl,python3,bash. OpenCode/OpenClaw additionally neednpm. - No Huawei Cloud IAM policies required — this skill operates on local bwrap sandboxes only.
参数确认 (Required Inputs)
| Parameter | Required | Description | Example |
|---|---|---|---|
--agent <name> | Yes (unless --all) | Target agent (see table above) | --agent opencode |
--all | Yes (unless --agent) | Operate on all 8 agents | --all |
--endpoint <url> | No | OpenViking server URL (default http://127.0.0.1:1933) | --endpoint http://192.168.1.100:1933 |
--api-key <key> | No | OpenViking API key (dev mode needs none). Never echo in chat | --api-key sk-xxx |
--dry-run | No | Show changes without applying | --dry-run |
--yes / -y | No | Skip authorization prompt (automation only) | --yes |
--json | No | status.sh: machine-readable output | --json |
核心命令
| 功能 | 命令 |
|---|---|
| 查看集成状态 | scripts/status.sh（--json 机器可读，--agent <name> 指定） |
| 验证 MCP 端点 | scripts/verify_mcp.sh |
| 集成单个 Agent | scripts/integrate.sh --agent <name> [--endpoint URL] [--api-key KEY] [--dry-run] [--yes] |
| 集成全部 Agent | scripts/integrate.sh --all |
| 解绑单个 Agent | scripts/unbind.sh --agent <name> [--dry-run] [--yes] |
| 解绑全部 Agent | scripts/unbind.sh --all |
Workflow
SKILL_DIR=/root/.agents/skills/huawei-cloud-openviking-agent-integration- Check status:
$SKILL_DIR/scripts/status.sh— per-agent state:template + live(active),template only(activates on restart),live only(lost on restart). - Verify MCP:
$SKILL_DIR/scripts/verify_mcp.sh— full MCP handshake (initialize → tools/list → health). - Integrate:
$SKILL_DIR/scripts/integrate.sh --agent <name>(or--all). - Unbind:
$SKILL_DIR/scripts/unbind.sh --agent <name>(or--all). - Rebuild OpenClaw:
stop + startvia job-env-manager API re-runsstart.sh. Scripts: references/related-commands.md.
Authorization & Safety
- Authorization is mandatory —
integrate.shandunbind.shrequire explicitconfirm(or--yesfor automation).--dry-runpreviews without authorization. - Do not fabricate integration state — always run
status.shto verify before reporting. - Never edit agent configs directly — all changes go through the skill scripts.
- No API keys in logs —
--api-keyvalues must never appear in output. - Every config modification creates a
.bak.<timestamp>backup for rollback.
Full rules: references/guardrails.md. Troubleshooting: references/troubleshooting.md.
Plugin Sources
The skill ships no plugin code — plugins are installed on demand at integrate time via domestic-first mirrors (Huawei Cloud npm → npmmirror → npmjs; GitHub raw mirrors for non-npm plugins). All downloads are byte-verified against GitHub blob SHA. Installed copies under /root/runtime/ are reused if upstream is unreachable (3-tier cache: sandbox node_modules → runtime cache → online).
References
| Document | Description |
|---|---|
| agent-configs.md | Per-agent config details, MCP tools, server info |
| guardrails.md | Safety, authorization, access permissions |
| troubleshooting.md | Failure scenarios and fixes |
| verification.md | Verification methods and acceptance criteria |
| related-commands.md | Restart/rebuild scripts, env vars |
Scripts (OO Architecture)
Base class (lib/base.sh) + registry (lib/registry.sh) + per-agent subclasses (agents/*.sh). Entry points (integrate/unbind/status.sh) are thin CLI parsers. All scripts are idempotent with .bak.<timestamp> backups. Adding a new agent = one file in agents/.
