Paperweight 0.7 adds MCP agent support

// 1h agoPRODUCT UPDATE

Paperweight 0.7 adds MCP agent support

Open-source, local-first desktop application Paperweight has released version 0.7.0, adding experimental AI agent access via a built-in local Model Context Protocol (MCP) stdio server. The feature allows connected AI agents to inspect mapped accounts and automate privacy workflows while restricting access to metadata and masking detected personal data.

// ANALYSIS

Hooking autonomous agents into private inboxes sounds like a security hazard, but Paperweight's local stdio MCP implementation demonstrates how privacy-conscious tools can safely adopt agentic workflows.

–Local-first safety: By communicating strictly over a local stdio MCP transport with zero cloud intermediaries, the integration maintains Paperweight's offline privacy posture when paired with local models like Ollama.
–Granular permission modes: Users can select between read-only access (for mapping exposures and querying senders) and read/write execution (enabling automated inbox cleanup and deletion notices).
–Metadata-only exposure: The server withholds raw message bodies and full headers, serving only app-level classifications and pre-masked personal data to minimize leak vectors.
–Autonomy risks: Automated privacy and deletion actions still rely heavily on the agent host's approval settings, posing risks if users configure unconstrained auto-running tools.

// TAGS

paperweightmcpagentprivacylocal-firstemailopen-source

DISCOVERED

1h ago

2026-09-21

PUBLISHED

2h ago

2026-09-21

RELEVANCE

7/ 10

AUTHOR

wslyvh

// KEEP READING

More AI developer news from the feed

EXPLORE FULL FEED

UPDATE43m ago

fal launches H3 Max Styles video presets

fal has introduced H3 Max Styles, bringing five specialized aesthetic presets directly into its fastest frontier video generation model, MiniMax H3 Max. The release enables creators to generate consistent stylized video—including 16-bit pixel art, low-poly 3D, and vintage VHS—without brittle prompt crafting at high inference speeds.

OPEN SOURCE1h ago

Mobile Verification Toolkit detects targeted mobile spyware

The Mobile Verification Toolkit (MVT) is an open-source forensic framework originally developed by Amnesty International Security Lab to detect traces of sophisticated mercenary spyware like Pegasus and Predator. Targeted at security researchers and investigative journalists, MVT automates forensic artifact extraction from iOS and Android devices, cross-referencing system telemetry against standardized STIX2 Indicators of Compromise to confirm targeted compromises.

UPDATE1h ago

Merge Gateway adds xAI Grok 4.7

Merge has announced that xAI's Grok 4.7 is now available on Merge Gateway, its unified API platform offering smart routing, fallbacks, caching, and spend visibility across AI models. The update emphasizes substantial gains in code generation and reasoning, with Grok 4.7 scoring 46.3% on CursorBench 4.0 compared to 40.4% for Grok 4.6, alongside a 71.0% mark on DeepSWE v1.1.