PtcRunner concepts
Use this page as a quick lookup for the names that appear in PtcRunner files, commands, and results. You can run the Quickstart without reading it first.
What do these names mean?
| Name | Meaning |
|---|---|
| Application | The code and configuration in ptc.json. |
| Project | A ptc-project.json file that remembers local paths and artifact settings. |
| Host file | A ptc-host.json file that installs credentials, models, tools, and outer limits. |
| Provider | A model, MCP server, replay fixture, or trace source installed under a stable name. |
| Alias | The stable provider name selected by ptc.json. |
| Workflow | Trusted PTC-Lisp code that coordinates the run, including model calls and agent policy. |
| Mission | The smaller environment where a model-authored program runs with selected data, components, and tools. |
| Component | One immutable PTC-Lisp module with a namespace and declared dependencies. |
| Prelude | A library of components shipped with PtcRunner, such as the replaceable agent.core loop. |
| Limits | Enforced ceilings for time, memory, calls, events, and result sizes. |
| Effect | A tool declaration of read or write; write tools require an explicit allow list. |
| Trace | The bounded operational record created by a run. It excludes prompts, responses, generated source, and tool payloads. |
| Private inspection | Optional sensitive records that may include the data excluded from a trace. |
| Command envelope | A stable JSON summary of a command's status, result or error, usage, and artifacts. |
The two main configuration files have different reach. ptc-host.json decides
what is available and holds credential references. ptc.json picks from that
installed set and can ask for less, but it cannot add credentials, endpoints,
commands, or wider limits. On a team, different people may maintain the files;
on one machine, you may maintain both.
Start with the Quickstart, then understand the generated project. The reference pages own the complete contracts.