Self-hosted workflows for coding-agent delivery. Orchestrate CodeBuddy, Codex and Claude Code with scripts and human approval gates. Includes Local workflows for coding assistants.
JavaScript agent-orchestrationagentic-workflowai-agentsai-codingclaude-codecode-reviewcodebuddycodexcoding-agentscursordeveloper-toolshuman-on-the-loopllmmcpmulti-agentsdlcself-hostedsopworkflow-automationworkflow-engine
"""Materialize MCP tools/list from a Selection. Used by UI and the generated server."""
from __future__ import annotations
from typing import Any
from app.catalog import TOOL_WARN_THRESHOLD, ToolSpec, tools_for_products
from app.models import BoundResources, Selection
def aliases_for(resources: BoundResources, kind: str) -> list[str]:
if kind == "redis":
return [r.alias for r in resources.redis if r.alias]
if kind == "cdb":
return [r.alias for r in resources.cdb if r.alias]
if kind == "tdsqlc":
return [r.alias for r in resources.tdsqlc if r.alias]
if kind == "tke":
return [r.alias for r in resources.tke if r.alias]
if kind == "cos":
return [r.alias for r in resources.cos if r.alias]
if kind in ("cls", "topics"):
out: list[str] = []
for logset in resources.cls:
for topic in logset.topics:
if topic.alias:
out.append(topic.alias)
return out
return []
def has_resources(resources: BoundResources, kind: str) -> bool:
return bool(aliases_for(resources, kind))
def tool_ready(tool: ToolSpec, products: list[str], resources: BoundResources) -> tuple[bool, str]:
if tool.requires_products:
missing = [p for p in tool.requires_products if p not in products]
if missing:
return False, f"需要同时启用产品: {', '.join(missing)}"
for kind in tool.requires_resources:
if not has_resources(resources, kind):
return False, f"缺少已绑定的 {kind} 资源,生成时不会注册"
if tool.name in ("k8s_events", "k8s_audit"):
topics = aliases_for(resources, "topics")
need = "events" if tool.name == "k8s_events" else "audit"
if need not in topics:
return False, f"需要绑定别名为 {need} 的 CLS 主题"
return True, ""
def input_schema(tool: ToolSpec, selection: Selection) -> dict[str, Any]:
props: dict[str, Any] = {}
required: list[str] = []
target_name = selection.project.target or "prod"
for param in tool.params:
schema: dict[str, Any] = {"type": param.type, "description": param.description}
if param.type == "array":
schema["items"] = {"type": param.items_type or "string"}
values: list[str] | None = None
if param.enum:
values = list(param.enum)
elif param.enum_from == "targets":
values = [target_name]
elif param.enum_from:
values = aliases_for(selection.resources, param.enum_from)
if values:
schema["enum"] = values
if param.default is not None:
schema["default"] = param.default
props[param.name] = schema
if param.required:
required.append(param.name)
out: dict[str, Any] = {"type": "object", "properties": props}
if required:
out["required"] = required
return out
def materialize(selection: Selection) -> dict[str, Any]:
"""Return tools/list-shaped payload plus grouping metadata for the UI."""
products = list(dict.fromkeys(selection.products))
disabled = set(selection.disabled_tools)
candidates = tools_for_products(products)
tools: list[dict[str, Any]] = []
skipped: list[dict[str, Any]] = []
registered: list[str] = []
for spec in candidates:
ready, reason = tool_ready(spec, products, selection.resources)
enabled = spec.name not in disabled
item = {
"name": spec.name,
"description": spec.description,
"product": spec.product,
"inputSchema": input_schema(spec, selection),
"annotations": {"readOnlyHint": spec.read_only},
"enabled": enabled,
"ready": ready,
"skipReason": "" if ready else reason,
}
if enabled and ready:
tools.append(item)
registered.append(spec.name)
else:
skipped.append(item)
grouped: dict[str, list[dict[str, Any]]] = {}
for item in tools + skipped:
grouped.setdefault(item["product"], []).append(item)
instructions = _instructions(products, registered)
return {
"instructions": instructions,
"tools": tools,
"skipped": skipped,
"grouped": grouped,
"registered": registered,
"counts": {
"products": len(products),
"resources": _resource_count(selection.resources),
"tools": len(tools),
"threshold": TOOL_WARN_THRESHOLD,
},
"warn": len(tools) > TOOL_WARN_THRESHOLD,
"warnMessage": (
f"将生成 {len(tools)} 个工具,超过 {TOOL_WARN_THRESHOLD} 容易让模型幻觉。"
"建议少选产品或在右侧关掉不常用工具。"
if len(tools) > TOOL_WARN_THRESHOLD
else ""
),
}
def _resource_count(resources: BoundResources) -> int:
n = (
len(resources.redis)
+ len(resources.cdb)
+ len(resources.tdsqlc)
+ len(resources.tke)
+ len(resources.cos)
)
for logset in resources.cls:
n += len(logset.topics)
return n
def _instructions(products: list[str], tools: list[str]) -> str:
lines = [
"腾讯云运维 MCP(只读云 API)。资源 ID 已绑定在 config,工具入参只用别名。",
"时间参数用 RFC3339。limit 超上限会报错,不会静默截断。",
]
if "tdsqlc" in products:
lines.append(
"TDSQL-C:库表用 db_*;慢 SQL 用 tdsqlc_slow_queries 看明细、"
"dbbrain_slow_top_sqls 看模板聚合。DBBrain Product 已固定为 cynosdb。"
)
if "cdb" in products:
lines.append("CDB:cdb_* 工具,DBBrain Product 已固定为 mysql。")
if "cls" in products:
lines.append("日志:cls_search_log / cls_histogram;topic 必须用已绑定别名。")
if "tke" in products:
lines.append(
"TKE:tke_list_pods + tke_pod_logs 看当前 Pod 日志;"
"tke_list_nodes / tke_list_addons 看节点和组件。"
)
if "tke" in products and "cls" in products:
lines.append("容器日志与 k8s 事件走 CLS 历史数据,不是实时 tail。")
if "redis" in products:
lines.append("Redis 只读云 API,不是 redis-cli 数据面。")
if "cos" in products:
lines.append("COS 只返回桶、对象列表和对象元数据,不下载对象内容。")
if tools:
lines.append("已注册工具: " + ", ".join(tools))
return "\n".join(lines)