Phase 7 Evidence Pack¶

Purpose¶

This evidence pack records the checks used to validate the Phase 7 namespace and dependency migration.

Scope Under Test¶

Phase 7 targeted the deepest remaining internal identity surfaces:

• Gradle root identity
• local java-tree-sitter project wiring
• vendoring java-tree-sitter into the parent repo so the new namespace commit is publishable
• buildSrc package namespace
• ai.serenade.treesitter Java imports
• JNI/header/native glue names for tree-sitter
• active build-script references that still depended on scripts/serenade
• direct sidecar imports of serenade-driver

Phase 7 did not target:

• stream-*.serenade.ai
• serenadecdn.com
• serenadeai/... repository ownership in external infrastructure config
• legacy website/blog/provenance content outside the live build/runtime path

Pre-Change Baseline¶

Check 1: Client Main Build¶

• Command: cd maestro/client && npm run build:main
• Result: passed.
• Observed warnings: optional ws native addon warnings for bufferutil and utf-8-validate.
• What it proved: the Electron main build was stable before Phase 7 edits.

Check 2: Core Compile Baseline¶

• Command: cd maestro && ./gradlew :core:compileJava -x downloadModels
• Result: failed before Phase 7.
• Observed failure: ModuleNotFoundError: No module named 'serenade' from core/bin/build-tree-sitter.py.
• What it proved: the inherited tree-sitter build path still depended on the old Python namespace and was a real Phase 7 blocker.

Post-Change Verification¶

Check 3: Client Main Build After Namespace Migration¶

• Command: cd maestro/client && npm run build:main
• Result: passed.
• Observed warnings: the same optional ws native addon warnings for bufferutil and utf-8-validate.
• What it proves: the sidecar wrapper and namespace changes did not break the Electron main build.

Check 4: Local Tree-Sitter Project Tests¶

• Command: cd maestro && ./gradlew :java-tree-sitter:test
• Result: passed.
• Observed tests:
• NodeTest > testGetChildren()
• ParserTest > testParse()
• TreeCursorTest > testWalk()
• What it proves: the renamed Java package, JNI headers, and native glue still work as a coherent local dependency.

Check 5: Core Tree-Sitter Build And Java Compile¶

• Command: cd maestro && ./gradlew :core:buildTreeSitter :core:compileJava :corpusgen:compileJava -x downloadModels
• Result: passed.
• Observed notes: existing ANTLR warning about a greedy block in CommandParser.g4.
• What it proves:
• core/bin/build-tree-sitter.py now resolves the Arqon Python namespace correctly
• core compiles against the local :java-tree-sitter project
• corpusgen now follows the renamed internal script paths

Check 6: Docs Build¶

• Command: mkdocs build
• Result: passed.
• What it proves: the Phase 7 tracking, evidence, and closeout docs publish cleanly.

Check 7: Targeted Residual Scan¶

• Command: rg -n "ai\\.serenade|serenade-driver|rootProject.name\\s*=\\s*\\\"serenade\\\"|scripts/serenade|import serenade\\.config|Java_ai_serenade_treesitter|ai_serenade_treesitter" maestro/{buildSrc,core,corpusgen,tree-sitter,client/src,client/static/custom-commands-server,replayer,config/Dockerfile,settings.gradle,README.md} -g '!**/node_modules/**'
• Result: only upstream serenade-driver package-manifest references and the local wrapper remained.
• What it proves: the live code/build surface no longer depends on inherited Java/JNI namespace names.

Evidence Summary¶

Phase 7 succeeded on the parts it was supposed to control:

• rootProject.name migrated to arqon-maestro
• core now depends on the local :java-tree-sitter project
• maestro/tree-sitter/java-tree-sitter is now vendored into the parent repo instead of remaining a gitlink to a local-only commit
• tree-sitter Java package moved from ai.serenade.treesitter to ai.arqon.maestro.treesitter
• JNI headers and native glue were renamed consistently
• buildSrc package migrated to ai.arqon.maestro
• active script-path references moved off scripts/serenade
• direct sidecar imports of serenade-driver were removed in favor of an Arqon-named wrapper

Residual Boundary¶

The following inherited names still exist, but they are outside the completed Phase 7 scope:

• upstream npm artifact names in package.json / package-lock.json
• external serenadeai/... repository ownership in config/languages.yaml
• inherited Docker image names and CDN/endpoint names
• legacy website/blog/provenance content

Conclusion¶

Phase 7 can be hard-closed because the internal namespace and dependency identity migration is complete on the live build/runtime path, and the remaining inherited names are now explicitly external or historical rather than internal technical identity leaks.