Register and parse a local SciFinder export file already visible to the server.

Copy a server-visible file into the upload area, then register and parse it.

Scan the NAS inbox for supported SciFinder exports and queue/register new files.

Return parse job status, stage, and error details.

List recent parse jobs, optionally filtered by status.

Return server health, configured paths, and indexed object counts.

Return the effective application config. Secrets are masked unless include_secrets is true.

Merge hot-reloadable application config updates into config.yaml and reload them.

Validate the current application config and report settings that require container restart.

Reload hot application config from config.yaml without restarting the container.

Search extracted reaction steps by text and structured condition filters.

Return one structured reaction step.

Return source text/page/parser provenance for a reaction step.

Clear extracted reactions for a document and parse it again.

Record DOI source verification performed by an agent or browser workflow.

Export extracted reaction steps as JSONL for manual labeling and regression checks.

Retry a failed or completed parse job by moving it back to the durable queue.

Retry recent failed parse jobs.

Generate embeddings for reaction steps using the configured embedding endpoint.

Return vector index coverage, model, and last error.

Search reaction steps semantically using the configured embedding endpoint.

Search the compound registry by name, CAS, SMILES, or InChIKey.

Return compound metadata, aliases, and linked reactions.

Merge source compound aliases/reaction links into a target compound.

Normalize a SMILES string when RDKit is available, then search compounds.

Send an image region to a configured MolScribe/DECIMER/OSRA-style endpoint and register candidate SMILES.

Compute regression metrics from a gold-set JSONL file.

Return the latest evaluation metrics.

Create a SQLite database backup, or report Postgres backup guidance.

Return NAS data/upload/evidence storage usage.

Clean generated evidence/cache files without deleting source documents.

Test one configured integration endpoint: llm, embedding, ocr, document_parser, structure_recognition, postgres.