{ "batch": "read1-013", "plugins": [ { "pr": 0, "name": "done-bell", "does": "Notifies users when Hermes finishes a turn or awaits approval, with a Desktop status chip and mute controls.", "useful": 4, "reliable": 3, "polish": 3, "creative": 2, "beyond": 4, "total": 20, "best": "Uses actual session and approval hooks plus Desktop busy state, so notifications do not depend on the model remembering to send them.", "worst": "Installation documentation uses a developer-specific Windows path and manual Desktop copying, and tests cover Python state rather than the Desktop notification path.", "standout": false, "blocked": null }, { "pr": 0, "name": "homeassistant", "does": "Connects Hermes to Home Assistant state-change events, device-control tools and notification delivery.", "useful": 4, "reliable": 5, "polish": 4, "creative": 3, "beyond": 5, "total": 26, "best": "A complete gateway adapter with reconnects, cron delivery, bounded REST calls and tested entity and dangerous-service controls.", "worst": "Requires an existing Home Assistant installation and powerful access token, and entity restrictions are opt-in rather than a default least-privilege setup.", "standout": false, "blocked": null }, { "pr": 152213, "name": "dome-relay", "does": "Queues structured cross-agent jobs through an HTTPS relay and retrieves or cancels their results in the originating conversation.", "useful": 3, "reliable": 4, "polish": 3, "creative": 3, "beyond": 4, "total": 21, "best": "Persistent job state, idempotency, compaction lineage and conservative uncertain-delivery handling are backed by focused relay and plugin tests.", "worst": "Users must operate a separate relay and worker plus an executive routine, and automatic push delivery is explicitly unsupported.", "standout": false, "blocked": null }, { "pr": 153824, "name": "hermes-janus", "does": "Snapshots workspace files into a content-addressed history and offers rewind, branching, merging and test-based branch selection.", "useful": 4, "reliable": 1, "polish": 2, "creative": 3, "beyond": 4, "total": 18, "best": "Automatic tool hooks and named restore points address a real coding-agent recovery problem without model calls.", "worst": "Branches reuse the same workspace, post-tool snapshots ignore failure results, and model-supplied tournament commands run directly with shell=True despite claims of isolated atomic recovery.", "standout": false, "blocked": null }, { "pr": 153629, "name": "uniichat-recall", "does": "Replaces Hermes memory with an append-only conversation log, summary tree and evidence-bearing recall tools.", "useful": 4, "reliable": 3, "polish": 3, "creative": 3, "beyond": 5, "total": 23, "best": "Integrates with the memory-provider lifecycle and returns original records with session IDs and excerpt offsets rather than only lossy summaries.", "worst": "Setup guidance is thin, Windows cannot load its fcntl store, and absolute zero-hallucination claims exceed a lexical matcher plus optional generated answers.", "standout": false, "blocked": null }, { "pr": 154018, "name": "artemis-owner-proof", "does": "Adds a mobile web companion that opens canonical Hermes Bot Chats through existing owner Web Access and gateway APIs.", "useful": 4, "reliable": 3, "polish": 4, "creative": 3, "beyond": 5, "total": 24, "best": "Readable bundled UI, scoped drafts, explicit reconnect logic and security-focused Markdown and identity tests make the mobile access concept concrete.", "worst": "Remote sign-in and canonical conversation continuity remain unverified in the disclosed evidence, and access retains full owner authority rather than a revocable phone-only grant.", "standout": false, "blocked": null }, { "pr": 153777, "name": "autonoma", "does": "Adds action-class trust scoring, extra approval prompts, an activity dashboard and deterministic handoff notes for unfinished sessions.", "useful": 4, "reliable": 3, "polish": 5, "creative": 4, "beyond": 5, "total": 26, "best": "Combines real approval hooks, correction-driven grant revocation and tool-log handoffs with a documented dashboard and extensive focused tests.", "worst": "The extra pre-tool gate catches internal exceptions and returns no directive, so its added protection can fail open; earned trust also rests on heuristic action classification.", "standout": false, "blocked": null }, { "pr": 153511, "name": "pocket-platform", "does": "Registers an authenticated radio-companion platform with bounded Hermes chat, notes and optional local music controls.", "useful": 2, "reliable": 3, "polish": 2, "creative": 4, "beyond": 5, "total": 21, "best": "Native session integration includes HMAC and peer checks, explicit owner enrollment, durable replay prevention and a dedicated-process veto on all model tools.", "worst": "The submitted install subdirectory is only the host plugin; usable hardware firmware, bridge service wiring, enrollment and a clean consumer installation remain external gaps.", "standout": false, "blocked": null }, { "pr": 153517, "name": "holdings", "does": "Adds a read-only Desktop portfolio and activity workspace with exchange and wallet connectors, evidence-linked chat and encrypted review storage.", "useful": 3, "reliable": 4, "polish": 5, "creative": 3, "beyond": 4, "total": 23, "best": "A substantial Desktop workflow combines original-record evidence, explicit private-model opt-in, human-confirmed transfer suggestions and security regression tests without moving funds.", "worst": "Protected reviews need an external vault secret and in-process loading, while live exchange matching and complete history coverage are explicitly limited.", "standout": false, "blocked": null }, { "pr": 149507, "name": "apple-health-mcp", "does": "Exposes local Apple Health export records, workouts and wellness summaries through a portable MCP package and workflow skill.", "useful": 3, "reliable": 3, "polish": 4, "creative": 2, "beyond": 3, "total": 18, "best": "The server source includes streaming export parsing, freshness and truncation reporting, privacy modes and meaningful parser and profile-boundary tests.", "worst": "The actual plugin launches an npm release rather than the pinned repository code, needs a first-run download and manual health exports, and adds no deeper Hermes lifecycle integration.", "standout": false, "blocked": null }, { "pr": 151159, "name": "fleet-sentinel", "does": "Probes remote Hermes gateway hosts using tailnet, TCP and SSH evidence, tracks outages and offers bounded recovery actions.", "useful": 3, "reliable": 3, "polish": 3, "creative": 3, "beyond": 3, "total": 18, "best": "Corroborated verdicts distinguish a live gateway socket from stale state files, with time-bounded probes and tests for misleading or malformed evidence.", "worst": "Recovery executes SSH mutations based on model-set dry_run and confirm flags rather than a Hermes approval prompt, and the PR description disagrees with the pinned implementation about storage and restart behavior.", "standout": false, "blocked": null }, { "pr": 152911, "name": "gmi-nexus", "does": "Adds GMI Cloud model discovery, task routing, benchmarking, multimodal generation and development tools with a dashboard and spend controls.", "useful": 4, "reliable": 3, "polish": 5, "creative": 3, "beyond": 4, "total": 23, "best": "Broad readable implementation combines real tool dispatch, scoped secrets, approval hooks, recovery tests and a documented demo and dashboard.", "worst": "BudgetGuard.daily_spend returns zero on storage errors, weakening the advertised hard daily cap, while goal planning largely uses fixed templates and clause splitting rather than general adaptive planning.", "standout": false, "blocked": null } ] }