Packages

ptc_runner

0.9.0
0.13.0 0.12.0 0.11.0 0.10.1 0.10.0 0.9.0 0.8.0 0.7.0 0.6.0 0.5.2 0.5.1 0.5.0 0.4.1 0.4.0 0.3.4 0.3.3 0.3.2 0.3.1 0.3.0 0.2.0 0.1.0

Secure BEAM sandbox runtime for LLM code mode and MCP aggregation. Run concurrent LLM/tool clients safely while agents orchestrate approved tools, call upstream MCP/OpenAPI servers, and transform data.

HexDocs

Current section

Files

Jump to
Readme
21 Versions
6 Dependencies
Dependants
Files
Activity
ptc_runner lib ptc_runner.ex
Raw

lib/ptc_runner.ex 

defmodule PtcRunner do

  @moduledoc """

  BEAM-native Programmatic Tool Calling (PTC) library.



  PtcRunner enables LLMs to write programs that orchestrate tools and transform

  data inside a sandboxed environment. The LLM reasons and generates code; the

  computation runs in an isolated interpreter with deterministic results.



  ## Core Components



  | Component | Purpose |

  |-----------|---------|

  | `PtcRunner.SubAgent` | Agentic loop: prompt → LLM → program → execute → repeat |

  | `PtcRunner.Lisp` | PTC-Lisp interpreter |

  | `PtcRunner.Sandbox` | Isolated execution with timeout/memory limits |

  | `PtcRunner.Context` | Tools and memory container |



  ## Example



      {:ok, step} = PtcRunner.SubAgent.run("What's 2 + 2?", llm: my_llm)

      step.return  #=> 4



  The SubAgent asks the LLM to write a program, executes it in the sandbox,

  and returns the result. See `PtcRunner.SubAgent.run/2` for all options.



  ## Guides



  - [Getting Started](guides/subagent-getting-started.md) - First SubAgent walkthrough

  - [Core Concepts](guides/subagent-concepts.md) - Context, memory, firewall convention

  - [Patterns](guides/subagent-patterns.md) - Composition and orchestration

  """

end