{ "batch": "read1-097", "plugins": [ { "pr": 0, "name": "done-bell", "does": "Adds end-of-turn notifications, approval-wait status, a status tool, and a small Desktop pane.", "useful": 4, "reliable": 4, "polish": 4, "creative": 3, "beyond": 5, "total": 25, "best": "It connects Hermes session and approval hooks to useful OS notifications and Desktop status, with tests and no credentials.", "worst": "Its value is a focused quality-of-life feature and its install instructions require manual Desktop-pane copying.", "standout": false, "blocked": null }, { "pr": 0, "name": "homeassistant", "does": "Connects Hermes to Home Assistant events and exposes tools to list, read, and control smart-home entities.", "useful": 4, "reliable": 4, "polish": 5, "creative": 3, "beyond": 5, "total": 26, "best": "A complete gateway platform and four device tools are backed by broad tests, detailed setup, scoped secrets, and configurable entity safety filters.", "worst": "Its broad device-control capabilities still depend on a powerful Home Assistant token, and users must configure its watch filters and network endpoint.", "standout": false, "blocked": null }, { "pr": 153402, "name": "enterprise-pathfinder", "does": "Provides SAP source guidance, live document reading, OData contract inspection and claim tests, with evidence-gated SAP skill writes.", "useful": 3, "reliable": 4, "polish": 4, "creative": 5, "beyond": 5, "total": 26, "best": "It makes Hermes's own research and learned SAP skills more verifiable through parsers, a local evidence ledger, and a strict write gate, with substantial tests.", "worst": "The first and only domain pack targets a specialized SAP audience, and the network reader can cache large remote documents locally.", "standout": true, "blocked": null }, { "pr": 153416, "name": "struggle-guard", "does": "Runs teacher-armed Socratic tutoring sessions, counts student attempts, limits hints by state, screens responses, and records a teacher trail.", "useful": 3, "reliable": 3, "polish": 4, "creative": 4, "beyond": 5, "total": 24, "best": "It uses Hermes middleware and persistent per-session state to make hint progression and teacher reporting more than prompt instructions, with adversarial and integration coverage.", "worst": "Its answer screening is heuristic and fail-open, and CLI streaming can reveal unscreened output unless disabled; only CLI has been verified.", "standout": false, "blocked": null }, { "pr": 151230, "name": "command-reviewer", "does": "Reviews shell and code execution calls using deterministic blocks, script-content inspection, model review, approvals, and hash-bound precedents.", "useful": 5, "reliable": 3, "polish": 4, "creative": 4, "beyond": 5, "total": 26, "best": "It adds a substantive defense layer to Hermes execution, including hard rules, review of actual script bytes, operator escalation, and extensive regression cases.", "worst": "It depends on private terminal internals and external model judgment, increasing maintenance and privacy risks while script contents are sent to the configured model.", "standout": true, "blocked": null }, { "pr": 152208, "name": "comemo", "does": "Replaces Hermes's single memory provider with parallel Honcho, Mem0, and Supermemory backends, merging reads and retrying failed writes.", "useful": 5, "reliable": 3, "polish": 4, "creative": 4, "beyond": 5, "total": 26, "best": "It changes Hermes's memory layer itself and adds multi-provider fan-out, deduplication, secret-scoped setup, and an offline retry queue.", "worst": "It forwards conversations to multiple configured cloud services and the repository has only a small provider test suite for a complex asynchronous integration.", "standout": true, "blocked": null }, { "pr": 149460, "name": "whoop-mcp", "does": "Packages an MCP server and workflow skill for reading WHOOP recovery, HRV, sleep, strain, and workout data through OAuth.", "useful": 3, "reliable": 2, "polish": 3, "creative": 2, "beyond": 3, "total": 16, "best": "The workflow is clear about consent, local setup, synthetic demo data, and the limitations of health guidance.", "worst": "The pinned plugin contains only manifests and a skill: its npx command downloads the actual server at runtime, which is outside the reviewed commit, and its PR flags an unresolved credential-policy question.", "standout": false, "blocked": null }, { "pr": 153257, "name": "conductor", "does": "Orchestrates local and remote coding agents, monitors runs without model polling, and reports completion, errors, or approval needs to the asking chat.", "useful": 5, "reliable": 4, "polish": 5, "creative": 5, "beyond": 5, "total": 29, "best": "It gives Hermes real multi-agent execution, persistent monitoring, safe launch recipes, remote support, and polished progress reporting, with a large behavioral test suite and real-run demo.", "worst": "It starts agents with the user's own permissions and imports a private Hermes session helper for routing, so the broad execution capability carries maintenance and operational risk.", "standout": true, "blocked": null }, { "pr": 150842, "name": "glasses-platform", "does": "Adds smart glasses as a Hermes gateway channel for voice turns, photos, streamed replies, and scheduled pushes.", "useful": 4, "reliable": 3, "polish": 4, "creative": 5, "beyond": 5, "total": 26, "best": "A real platform integration brings Hermes sessions, memory, approvals, photos, and later pushes to wearable devices, with extensive simulator and gateway tests.", "worst": "A device token grants full agent access, the default listener binds all interfaces without TLS termination, and real glasses hardware remains untested.", "standout": true, "blocked": null }, { "pr": 150500, "name": "hermes-windows-firewall", "does": "Adds read-only tools that explain Windows Firewall rules and profile settings directly from the Windows registry.", "useful": 3, "reliable": 4, "polish": 4, "creative": 3, "beyond": 3, "total": 20, "best": "Its locale-independent registry parser preserves unknown fields, distinguishes defaults from observed values, and has unit and live Windows tests.", "worst": "It solves a narrow Windows-only diagnostic task and does not change Hermes's platform or core capabilities.", "standout": false, "blocked": null }, { "pr": 153976, "name": "model-kombat", "does": "Runs structured head-to-head Hermes model evaluations, scores answers mechanically, tracks lane-specific Elo, and presents results in Desktop and kanban.", "useful": 4, "reliable": 4, "polish": 5, "creative": 5, "beyond": 5, "total": 28, "best": "It combines adaptive task selection, mechanical scoring, trend memory, capped pure-fight spend, and a polished live Desktop scoreboard, backed by extensive tests.", "worst": "The optional code scorer runs model-written code as the user's account and explicitly is not a security sandbox; agentic billing is also uncapped.", "standout": true, "blocked": null }, { "pr": 154007, "name": "swarm-studio", "does": "Uses a planner and parallel model agents to build multi-page websites, apps, games, promotional media, and deployable prototypes on an interactive Desktop canvas.", "useful": 4, "reliable": 3, "polish": 5, "creative": 5, "beyond": 5, "total": 27, "best": "It ships an unusually complete creative product with live multi-agent canvas, Desktop integration, publishing and game/app workflows, extensive docs, and a large test suite.", "worst": "Its enormous vendored codebase, CAUTION install scan, external model and publishing services, and multi-gigabyte first app build create meaningful reliability and setup friction.", "standout": true, "blocked": null } ] }