Run agent tool calls through pctx Code Mode and unified MCP
Put pctx between agents and tools so MCP servers and custom tools can run through a code-oriented execution layer instead of one-off tool calls.
npx skills add agentskillexchange/skills --skill run-agent-tool-calls-through-pctx-code-mode-and-unified-mcp
Use pctx when an agent needs to combine several tools or MCP servers into token-efficient code-mode workflows. The operator starts pctx, registers upstream MCP servers or Python tools, and exposes a unified interface that lets supported agents run tool-heavy work through executable code rather than repeatedly negotiating individual tool calls. This is best for agent builds where tool orchestration, sandboxed execution, and compact context handoff matter more than a plain MCP server list. The boundary is Code Mode and unified MCP tool execution for agents, not a generic CLI, SDK, or hosted integration platform.
What this skill actually does
Inputs and prerequisites: pctx CLI, Python SDK or npm package, MCP-compatible client or OpenAI Agents SDK.
Setup notes: Install with Homebrew using brew install portofcontext/tap/pctx or npm using npm i -g @portofcontext/pctx. Start Code Mode with pctx start, or initialize a unified MCP setup with pctx mcp init, add upstream servers with pctx mcp add, and run it with pctx mcp dev or pctx mcp start –stdio.
Source and verification boundary: use https://github.com/portofcontext/pctx as the canonical reference before running the workflow; keep commands, API calls, CLI usage, and generated outputs reviewable against that upstream source.
Framework fit: publish this as a MCP workflow only when the operator can invoke the documented toolchain directly, rather than treating the upstream project as a generic product listing.