Ingest Reports

Run report files from docs/database/CTI through extraction, quality checks and scoring, into the active database. Reports are processed one at a time; a report is only written to the database once it has been fully processed, so cancelling a run leaves nothing half-ingested. New files can be added on the Report Files page.

Engine

LLM enhanced checking... Regex candidates verified by a local Ollama model, plus document scope classification. Most accurate; minutes per report. Python only Same extraction, verified with local heuristics only. No LLM needed; seconds per report.

Log detail

Download from URL

Fetch a report PDF from the web into docs/database/CTI, then queue it for ingest with the engine and log detail above. The file must be a PDF; links to web pages are refused. Only public http(s) addresses on the standard ports are allowed - local and private network addresses are blocked, including via redirects.

Not yet ingested

	Filename	Size

Queue

Follow running log