List the available topic profiles by name (e.g. hospital-tech).

A profile defines which subreddits to mine, the search terms, and the pain lexicon. Use this first to discover what can be swept or queried.

Return one profile's full configuration as a dict.

Includes its subreddits, search_terms, pain_lexicon (weighted regex rules), pain_threshold and the evidence thresholds that gate reported gaps.

Run a full two-stage sweep for a profile and persist the results.

Stage 1 pulls + scores posts across the profile's subreddits and search terms; stage 2 deep-fetches comments on the highest-pain threads. This makes live Reddit calls and may take a while. time (hour|day|week|month|year|all), limit (posts per sub) and max_threads override the profile defaults when non-empty/non-zero. Returns a SweepResult summary (counts, per-subreddit tallies, top matched patterns).

Search Reddit live for a query and store the matching posts.

If subreddits is given the search is restricted to each of them (results merged + de-duplicated); otherwise it is a site-wide search. Posts are stored so you can drill into them later with reddit_query / reddit_get_evidence / reddit_fetch_thread. Returns the stored posts as dicts (id, subreddit, author, score, permalink, title, body, ...).

Fetch a post's comment tree live and store the comments.

post_id may be a bare id (abc123) or a fullname (t3_abc123). Returns {"post": <post dict or None>, "comments": [<comment dicts>], "comments_collected": N}. The post object is included when it is already in the store (e.g. from a prior sweep/search). Comments are stored so you can cite them as evidence afterwards.

Query already-stored items (no network).

Filters: profile (which sweep collected it), subreddit, min_pain (lower bound on the deterministic pain score), contains (case-insensitive substring over title+body). sort is one of pain|score|new|comments. Returns hydrated item dicts including their matched lexicon patterns.

Resolve stored items into citable evidence for the given ids.

Returns one EvidenceItem dict per known id (in the order requested) with the permalink and a trimmed verbatim quote — use this to ground any claim in a real Reddit link instead of paraphrasing. Unknown ids are skipped.

Summary stats for the store (optionally scoped to one profile).

Returns totals, the post/comment split, a per-subreddit breakdown, the most frequent matched patterns, and the (min, max) created-utc date range.

Export stored items as a serialized string in json, csv or md.

Applies the same profile / min_pain filters as reddit_query and returns the rendered payload as text (highest pain first). Useful for handing a dataset back to the user.