Menu
Use control-plane MCP to discover and call connected tools or run account workflows.
The review-first prompt is selected by default. You can switch to a direct command or download a local copy.
Read SKILL.md and any companion files shown by SkillsMP before deciding whether to install.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
A direct command skips the review prompt. Inspect the source before running it.
npx skills add https://github.com/tangle-network/skills --skill tangle-control-planeThe command stays on one line. Scroll horizontally to inspect it before copying.
Prefer a local copy? Download the files currently available to SkillsMP.
Showing SKILL.md
Based on SOC occupation classification
| name | tangle-control-plane |
| description | Use control-plane MCP to discover and call connected tools or run account workflows. |
This skill assumes the tangle-control-plane MCP server is connected (this plugin registers it). It exposes your Tangle account as tools: connected integrations, workflows, skills, usage/billing, and account-level sandbox management.
Do not memorize a tool catalog from this file — it drifts. The live sources of truth are the tools themselves (get_workflow_schema, search_integration_actions, describe_integration_action) and GET {TANGLE_HUB_URL}/mcp/info (public: current scopes + tool list + health). This skill teaches the loops and the footguns, not the catalog.
Never guess an integration action path or a model slug. Discover it from a tool first. Guessed paths and invented model slugs are the two most common failures.
list_integrations — what the user has connected.search_integration_actions — find candidate actions by intent.describe_integration_action — get the exact path + input schema.invoke_integration — call it with the discovered path and args.Use this for a single action ("email me the summary", "open a GitHub issue"). For anything recurring, author a workflow instead (Loop B).
get_workflow_schema — the deployment's real authoring capabilities and model catalog. Read it before writing any YAML.validate_workflow — dry-run compile the YAML. Fix every reported error.create_workflow — save it.run_workflow — enqueue; returns { runId }.get_workflow_run — poll with runId until status is terminal (succeeded / failed / cancelled).list_workflow_runs — recent history.author_workflow does create-workflow + mint-skills in one call, but requires the admin scope.
agent.run.model is passed to the router verbatim; an invented or date-suffixed slug passes validate_workflow and then hangs at runtime. Copy an exact provider/model slug from get_workflow_schema's model list, or omit model for the default. Never invent one.github.issues.create, not createIssue. Never guess — search/describe first.create_workflow on YAML you have not run through validate_workflow.The server authenticates with TANGLE_API_KEY (an sk-tan-… key) sent as Authorization: Bearer. Tools are gated by the key's scopes:
| Scope | Grants |
|---|---|
read | list / get / validate workflows, runs, integrations, usage |
workflows:write | create / update / run workflows |
integrations:invoke | invoke_integration |
admin | wildcard, incl. author_workflow + key management |
If a call fails with a scope/auth error, tell the user exactly which scope to add. The key must be user-owned (team keys are rejected). invoke_integration also needs a positive account balance, and the endpoint rate-limits at 120 requests/min per key.
If /mcp shows the server failing or the tools are missing:
curl "${TANGLE_HUB_URL:-https://id.tangle.tools}/mcp/info" — JSON back = endpoint live; 404 = the deployment has not enabled it (CONTROL_PLANE_MCP_ENABLED).TANGLE_API_KEY is exported in the shell that launched this agent — it expands at connection time; if unset, the Bearer header is empty and auth fails.See this plugin's README for one-time setup.