lossless-group_perplexed-plugin/release-notes/0.2.0.md

Ask Gemini joins Perplexity, Claude, Perplexica, and LM Studio as Perplexed's fourth research provider — Google Search grounded, with the per-claim citation attribution Claude's web_search round-trip silently loses.

Why care?

If you fill research notes in Obsidian — concept profiles, source dossiers, vocabulary entries, anything where the citation matters as much as the body — Perplexed 0.2.0 changes two things that show up the moment you use it.

You have a fourth research provider, and it cites better than the three before it. Ask Gemini joins the lineup with full Google Search grounding turned on by default. Where Claude's newer web_search_20260209 drops per-claim citations on the round-trip through its dynamic-filtering sandbox (sources come back, attribution doesn't — text blocks return with citations: null), Gemini's groundingSupports[] carries the segment-to-source mapping intact. So the ### Citations footer that lands in your note is actually attached to the prose above it, with the verbatim quote per source.

Your shared editorial rules stop being duplicated across templates. Four directory templates (concept, vocabulary, source, toolkit) used to repeat the same mermaid-discipline / citation-enforcement / image-placement / research-framing guidance inside each one, with some of it hardcoded in TypeScript users couldn't touch. Now those rules live in vault-visible partials/ and preambles/ folders. Fix the mermaid rules once in partials/mermaid-discipline.md and every template picks the fix up on the next generation.

The full provider lineup as of 0.2.0:

Provider	Endpoint	Cost	Best for
Perplexity	api.perplexity.ai	Paid	Source-cited research with consistent citation formatting
Anthropic Claude	api.anthropic.com	Paid	Longer-form reasoning with web_search and adaptive thinking
Google Gemini (new)	generativelanguage.googleapis.com	Free tier (no credit card) + paid	Google Search grounding with per-segment citation attribution
Perplexica / Vane	localhost:3030 (self-hosted)	Free	Privacy-sensitive research; runs entirely on your machine
LM Studio	localhost:1234 (local)	Free	Local-only inference, no network

Open Obsidian → Settings → Community Plugins → Browse → "Perplexed" → Install → Enable, then paste a Gemini API key from aistudio.google.com (free tier, no credit card required) into Settings → Perplexed → Gemini (Google).

Ask Gemini — what's in the box

A new Ask Gemini command opens a modal mirroring the Ask Claude shape: question textarea, model dropdown, behavior toggles, Cmd-Enter to submit. Default model is gemini-flash-latest (Google's always-current Flash alias, free-tier friendly) — gemini-pro-latest and pinned gemini-2.5-pro / gemini-2.5-flash are also selectable.

When google_search grounding is enabled (default on), the plugin streams the answer into your note and parses two layers of provenance from the response:

Layer	Field	What it gives you
Page-level	groundingChunks[]	URL + title per page Gemini consulted
Segment-level	groundingSupports[]	Text span (segment.text) → indices into groundingChunks[]

The plugin walks both layers and merges them by URL: page-level chunks become URL-and-title fallbacks; segment-level supports enrich them with the verbatim quote Gemini grounded against. Citation footer mirrors Claude's exactly ([N]: [Title](url). > cited_text) so Cite Wide's hex-substitution pass works on Gemini output too.

The two Gemini-specific quirks Perplexed handles for you

Anyone who's wired the raw Gemini API will recognize these:

1. chunk.web.uri is a short-lived Google redirect. The URL Google returns is a vertexaisearch.cloud.google.com/grounding-api-redirect/AUZIY… redirect that expires roughly 30 days after the response. Naive "just write the URL we got" means every cited note rots on a clock. Perplexed resolves each redirect via Obsidian's requestUrl (Node-side, no CORS friction) and parses the destination page's <link rel="canonical"> / <meta property="og:url"> for the durable source URL — and parses <meta property="og:title"> / <title> for the real page title. So a citation that would naively render as [bloomberg.com](vertexaisearch.cloud.google.com/...) instead lands as [Bloomberg Beta — Bloomberg LP](https://www.bloomberg.com/company/bloomberg-beta/).

2. searchEntryPoint.renderedContent is 5KB of inline-styled HTML. Google's grounding terms ask that this Search Suggestions chip be displayed when google_search is used; the chip itself is <style>-tag-prefixed HTML that Obsidian's Markdown renderer strips, leaving orphan <div> shells. Perplexed substitutes a Markdown-native ### Google Searches section: a bullet list of the queries Gemini actually issued, each linked to google.com/search?q=… so you can re-run the suggested search. Same end-user behavior, rendered in a form Obsidian can actually display.

Partials + preambles — shared template guidance, vault-visible

The four directory templates (concept-profile, vocabulary-profile, source-profile, toolkit-profile) historically duplicated their mermaid-discipline / citation-enforcement / image-placement / research-framing guidance inside each template, with some of it hardcoded in TypeScript constants users couldn't edit. 0.2.0 moves that guidance into two peer folders alongside templates/:

Content-Dev/
├── templates/         (the four profile templates)
├── partials/          (NEW — referenced from templates via {{include: name}})
│   └── mermaid-discipline.md
└── preambles/         (NEW — auto-attached to every Perplexity request)
    ├── inline-citation.md       (was a hardcoded TS constant)
    ├── image-placement.md       (was a hardcoded TS constant)
    └── research-framing.md      (was a TS function)

{{include: name}} resolves recursively (depth-limited, cycle-detected). Preambles auto-attach to every request with bundled defaults as fallback, and per-template overrides live in the cft fence: preambles: { system: [...], skip-user: [...], skip-all: true }.

What else this release brings

• Eight previously-hidden settings sections render again. A long-standing addClass('perplexed-json-textarea is-tall') bug at two DOM API call sites was throwing InvalidCharacterError (the method takes a single token, not a space-separated string), aborting display() mid-render and silently hiding every settings row after the Article Generator block. Same shape of fix as the May 19 batch that switched eight activeDocument.createEl calls to containerEl.createEl.
• System-prompt textareas widened to full-width 3-line rows. The three system-prompt settings used Setting.addTextArea which crammed them into the 200px right edge of the row at 2 lines tall. Now each gets a name+description row on top and a full-width resizable textarea below.
• Template seeder idempotent. Both "Folder already exists" and "File already exists" race conditions (Obsidian's in-memory file index lagging the adapter write) now get swallowed cleanly. No more red console lines on plugin load.
• ERR_NETWORK_CHANGED becomes a useful sentence. WiFi roam, VPN reconnect, sleep/wake events mid-Perplexity-stream now write "Network changed mid-stream — re-run the query" into the note instead of a 30-line stack trace.
• Ask Gemini modal with Google-flavored visual identity: gradient title text (blue → purple → red → yellow via background-clip: text), Google four-color hairline under the header, Gemini-blue focus ring on the prompt textarea, gradient-on-hover CTA. Theme tokens everywhere structural so light, dark, and community themes all carry through.

Verify the release assets

Every release asset in 0.2.0 is signed via GitHub's artifact attestation system at build time (the workflow introduced in 0.1.2). Before manual install, verify provenance:

gh attestation verify main.js --repo lossless-group/perplexed-plugin
gh attestation verify manifest.json --repo lossless-group/perplexed-plugin
gh attestation verify styles.css --repo lossless-group/perplexed-plugin

Exit code 0 means the file came out of a legitimate build of the source repo.

The Lossless Group plugin family

Plugin	What it does
Cite Wide	Stable hex-identifier citations + URL-based dedupe + LLM-paste research conversion
Image Gin	Recraft + Ideogram image generation, Magnific stock search, ImageKit CDN, drag-gate
Perplexed (this release)	Source-cited research from Perplexity, Claude, Gemini, Perplexica, or LM Studio + per-directory templates
Metafetch	Pull OpenGraph metadata into note frontmatter via OpenGraph.io or Microlink

Together they form a research-to-publication pipeline: Perplexed generates source-cited prose, Cite Wide canonicalizes the citations to stable hex identifiers, Metafetch fills URL frontmatter with OG metadata, Image Gin generates matching imagery.

If Perplexed saves you time, buy us a coffee. The plugins are free; the coffee keeps the next one coming.

Under the hood — full per-day changelog index

Each substantive piece of work shipped in 0.2.0 has its own per-day changelog in the repo:

• changelog/2026-05-19_01.md — partials + preambles paradigm; the one-line fix that resurrected 11 hidden settings sections
• changelog/2026-05-19_02.md — Gemini provider lands; response shape dissected with curl
• changelog/2026-05-20_01.md — Obsidian-shipping pass: CORS-blocked fetch → requestUrl + canonical/og:url + <title> parsing; ?key= → X-goog-api-key header; default model → gemini-flash-latest; Gemini modal CSS; safeCreateFile race fix; ERR_NETWORK_CHANGED UX; system-prompt textareas promoted; addClass token bug fixed at two more call sites

Full marketing-grade release narrative with deeper context: changelog/releases/0.2.0.md.