Engli Explorer

Memory Store

Store a memory (fact, decision, preference, note, todo) in a project. Memories persist across sessions and are recalled semantically. Use a project per codebase/topic for isolation. Alternatively pass `file` instead of `content` to ingest a whole file — it is chunked into recallable memories AND the original is archived (when a bucket is configured).

Memory Recall

Semantically recall stored memories. Searches one project, or every project in your namespace when project is omitted.

Memory Ingest File

Ingest a whole file into project memory. Chunks along the format's natural seams — markdown by headings, code by definitions, CSV by rows, JSON/JSONL by entries, prose by paragraphs — then every chunk is recallable. Binary files work too: pass the raw bytes base64-encoded with encoding:'base64' — PDF, Word (docx/doc), Excel (xlsx/xls), PowerPoint (pptx/ppt), OpenDocument (odt/ods/odp), EPUB, RTF, email (.eml), Jupyter notebooks, zip/gz archives (each document inside is extracted), and images (pn

Memory Add Link

Add a web link to a project: the page is fetched and scraped server-side into recallable memory, and the URL itself is kept as a link memory. Use when the user shares a URL that belongs to a project, or says 'add/scrape this link'.

Memory Index History

Index a conversation's or project's WHOLE HISTORY into memory in one call. Use when the user says 'save/index this conversation/project into memoception'. Best: distill the history into entries — one per decision, fact, artifact, or milestone, each with a clear title — and pass them in `entries`. Or pass the raw `transcript` and it will be chunked along its seams. Safe to re-run: entries that near-duplicate existing memories are skipped, so incremental re-indexing only adds what's new.

Memory Projects

List this namespace's projects with memory counts and last activity.

Memory Forget

Delete memories by ids, by project, or by source filename. Provide at least one filter.

Memory Consolidate

Consolidate a project: near-duplicate memories merge into one stronger memory (its weight accumulates and boosts future recall). Run this periodically so long-lived projects compress instead of silting up.

Memory Export Training

Export a project's memories as training JSONL — prompt/completion pairs (or ChatML messages). Turns accumulated memory into fine-tuning data.

Grove Search

Search the shared Engli Grove knowledge graph (studies, tools, translations, protocols…) — global verified knowledge, read-only.

Research And Store

Research a subject online and store the findings into a project in one call. Discovers sources (web search when a subject is given), reaches each page through a clean-extraction reader with fallback routing, chunks it into recallable memory, keeps each URL as a link memory, and writes a research-summary note. Pass explicit `sources` (URLs), a `subject` to search for, or both. Best-effort per source — one dead link never fails the call.