Skip to content

feat(validation): persist eligible_files.json for post-index validation#69

Open
Dead-Bytes wants to merge 2 commits into
pre-releasefrom
feat/validation
Open

feat(validation): persist eligible_files.json for post-index validation#69
Dead-Bytes wants to merge 2 commits into
pre-releasefrom
feat/validation

Conversation

@Dead-Bytes
Copy link
Copy Markdown
Collaborator

Summary

  • Adds writeEligibleFiles to the flat-folder strategy: persists .bytebell/eligible_files.json (knowledgeId, commitHash, sorted file list, affected folders) to the source layer between scan-and-classify and the parallel analyse-small / analyse-big phases.
  • Writes to local disk when source.localRepoDir is set and to archiveSink when configured; hard-fails if neither target is available. Read back by @bytebell/knowledge-validation to cross-check that every file the analyzer was asked to process landed in Neo4j.
  • Brings pre-release up to current main: drops the retired db, db-core, graph-db, graph-core, and sqlite packages, consolidates @bytebell/config schema into a single file, removes the mongo / neo4j provider.ts shims and the DbProgressReporter, and prunes unused exports under @bytebell/types (analysis, config, database, graph).
  • Server, MCP, and queue packages updated for the new import surfaces; dependency lists trimmed accordingly.

Test plan

  • bun run typecheck is clean.
  • Run a GitHub ingestion end-to-end against a small repo; confirm .bytebell/eligible_files.json appears under source.localRepoDir before any small/big file analysis log line.
  • Run the same ingestion with only archiveSink configured (no local repo dir); confirm the artifact is pushed through the sink and analysis proceeds.
  • Run with neither target available; confirm the strategy fails fast with the expected error rather than producing an un-validatable knowledge.
  • Smoke-test @bytebell/knowledge-validation against a freshly-indexed knowledge and confirm it reads the artifact and reports parity.
  • Boot @bytebell/server and hit /health, /repos, /stats, /mcp-stats to confirm the package removals did not break route wiring.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants