Skip to content

Release v3.0.0

Choose a tag to compare

@github-actions github-actions released this 01 Nov 18:46
9fa1101

Prologue now has a web interface, the Prologue Workbench, served by the Receiver. It takes a prologue from defining its systems, through discovering what the code does and watching captures arrive, to reviewed event suggestions, an event model you can export as Screenplay, and a tracked rebuild. The Receiver becomes an Arc and Chronicle application for this, so it now needs a Chronicle kernel. See Running the Workbench.

Added

  • The Prologue Workbench at / on the Receiver, with seven steps (Systems, Discover, Capture, Explore, Events, Model, Rebuild) in the Cratis design system, a prologue switcher and a phone layout. Discover, Capture and Rebuild are optional; at least one of Discover and Capture stays on (#65) (#67)
  • docker compose -f Source/Receiver/docker-compose.yml up --build runs the Workbench with the Chronicle kernel it needs (#66)
  • Systems: define a prologue's systems, each with a repository host, repository, branch and sub path, so several systems can share one monorepo; pick runtime sources, see what is and is never recorded, and copy the generated cratis-prologue.json, which points the Extractor at the Receiver you are using (#68)
  • Repository connections for GitHub, Azure DevOps and GitLab, plus local folders; system repositories are cloned read-only, shallow and sparse into Prologue:Repositories:CloneDirectory, with one clone shared by systems in the same repository (#69)
  • AI setup through Cratis.AI: providers (Anthropic, OpenAI, Azure OpenAI, OpenAI-compatible for models in your network, Z.ai, OpenAI Codex, GitHub Copilot), pools with tier, ordering and fallback, and Prologue's agents (Dramaturg, Prompter, Usher, Interpreter). The shell shows whether code stays in your network: Local only, Local first or Leaves network (#70)
  • Discover analyzes a system's source, or any folder, into an application map of capabilities, decision points with file and line, and dependencies between systems, matched against what the captures show; maps can be shared as a read-only link. ASP.NET controllers are mapped even without an AI provider (#71)
  • Capture shows captures as they arrive, with rate, command shapes, tables touched, trace-linked share, uncorrelated writes, coverage by route, a timeline per capture, and an explanation of each accepted or refused decision (#72)
  • Scenario scripts: mark the business scenario you are running, and the captures that arrive meanwhile carry its label into interpretation (#73)
  • Explore shows a command-by-table matrix with how often each write happens, observations per shape, journeys inferred from ids in request paths, uncorrelated writes by hour, and questions about the captures answered with cited shapes. Shapes can be excluded from interpretation or split by response status (#74)
  • Business events with reasons, for example "Loan refused, because no copy was available (LoansController.cs:71)", using the rule Discover found in code when there is one (#75)
  • Events runs interpretation from the Workbench and lists suggested events with evidence from captures, scenarios, decisions and code. Suggestions can be accepted, renamed, rejected and restored, and the decisions survive re-runs. Interpreter questions, consistency boundaries and columns that are not events are shown alongside (#76)
  • Model shows the accepted event model as a board of slices, generates its Screenplay .play, served as a download from GET /prologues/{prologueId}/model.play, and turns accepted slices into Rebuild slices (#77)
  • Rebuild tracks each slice from observed to cut over. The Extractor can shadow commands to the new system through prologue.shadowing, without changing the caller's response, and the Receiver accepts parity on POST /prologues/{prologueId}/parity. Differences are shown side by side, a slice cannot be cut over while differences are unresolved, and new command shapes or columns after the model is accepted show up as drift. See Shadowing (#78)
  • ${secret:<name>} placeholders in cratis-prologue.json resolve from Prologue:Secrets:<name> or the environment variable PROLOGUE_SECRET_<NAME>; a secret that cannot be resolved stops startup with its name in the message
  • Optional Prologue:Discover:AllowedRoots limits the folders Discover may read
  • Documentation for running the Workbench, a walkthrough of the seven steps, and reference pages for the Receiver's configuration, repository connections, AI setup and shadowing (#79)

Changed

  • The Receiver now requires a Chronicle kernel and reads its read models from the same MongoDB replica set Chronicle writes to. Set Cratis:Chronicle:ConnectionString and Cratis:MongoDB:Server before upgrading; the docker compose file shows a working setup
  • The Receiver and Extractor images run as the non-root user app (uid 1654). Make mounted volumes, such as the Extractor's capture folder and the Receiver's clone directory, writable by that user
  • ICaptureStore.Store in Cratis.Prologue.Storage returns Task<bool>, true when the capture was stored now; storing a capture that is already stored, as a retried post does, is no longer an error. Implementations of the interface need the new return type
  • The capture endpoints answer as soon as a capture is stored and record it for the Workbench in the background, so a slow or unavailable Chronicle never holds up the Extractor
  • IInterpreterSessionFactory.CreateNew in Cratis.Prologue.Interpretation takes an optional additionalEvidence for refinement, and InterpreterSessionState gains Renames. Recompile code that implements or calls these against the new package

Security

  • Repository tokens are sent only to the host of their own connection, over https; git runs with only https allowed and without system or global git configuration, and local or file:// repositories are refused for hosted connections
  • Shadowed requests carry only an allowlist of headers plus the ones a rule names, so credentials and identity headers are not forwarded