{ "batch": "read2-006", "plugins": [ { "pr": 0, "name": "done-bell", "does": "Notifies the user when a Hermes turn finishes or waits for approval, with a desktop status chip, optional chime, and mute controls.", "useful": 3, "reliable": 3, "polish": 3, "creative": 2, "beyond": 4, "total": 19, "best": "Hooks and a desktop surface make turn completion visible while the user is away, with tests for its event and storage paths.", "worst": "The use case is helpful but narrow, and the README describes a manual Windows pane-copy step alongside a development install.", "standout": false, "blocked": null }, { "pr": 0, "name": "homeassistant", "does": "Connects Hermes to Home Assistant events and device APIs, providing four tools for querying and controlling smart-home entities plus gateway and cron delivery.", "useful": 5, "reliable": 4, "polish": 4, "creative": 3, "beyond": 5, "total": 26, "best": "It gives Hermes a substantial, native smart-home integration across tools, event routing, and notification delivery.", "worst": "Device control can have real-world effects; the plugin relies on a user-supplied long-lived token and its README documents filtering, but no screenshots or test details were inspected in this pass.", "standout": false, "blocked": null }, { "pr": 153240, "name": "flowcanvas", "does": "Provides Hermes Desktop with a visual node canvas for building, editing, and running reusable workflows across Hermes tools, MCP, models, and media.", "useful": 4, "reliable": 4, "polish": 5, "creative": 5, "beyond": 5, "total": 28, "best": "A tested, polished Desktop workflow runtime adds graph execution, caching, bounded loops, media previews, and agent-directed graph editing.", "worst": "Its breadth raises maintenance and integration risk, and flows that use external providers require separately configured services and may incur provider costs.", "standout": true, "blocked": null }, { "pr": 151411, "name": "salts-secrets", "does": "Routes non-owner gateway traffic to a separate public Hermes profile, limiting guests to owner-approved context and tools, with dashboard publishing and revocation controls.", "useful": 4, "reliable": 3, "polish": 5, "creative": 5, "beyond": 5, "total": 27, "best": "Identity-based separation using Hermes profile routing, a guest-specific tool gate, an owner dashboard, and a working demo tackles a difficult privacy boundary.", "worst": "This changes gateway routing, profiles, sessions, and API access in a security-critical way; the extensive custom integration still has substantial compatibility and operational risk.", "standout": true, "blocked": null }, { "pr": 152960, "name": "pwn-board", "does": "Tracks open-ended security or debugging investigations as evidence-backed intents, findings, and goal tests, with operator review and a Desktop pane.", "useful": 3, "reliable": 4, "polish": 4, "creative": 4, "beyond": 4, "total": 23, "best": "A structured evidence ledger and independent verification can make long investigations auditable and reduce unsupported agent conclusions.", "worst": "The workflow is specialized and the goal test runs a configured command through Hermes terminal, so setup and execution policy require care.", "standout": false, "blocked": null }, { "pr": 153853, "name": "keryx", "does": "Lets Hermes conduct an outbound call through the user’s paired Linux phone, after per-call approval, using realtime speech models and returning a transcript and outcome.", "useful": 3, "reliable": 4, "polish": 4, "creative": 5, "beyond": 5, "total": 26, "best": "It brings live phone conversations into Hermes through a real device integration, and enforces approval, disclosure, blocked numbers, rate limits, and a call cap in code.", "worst": "Linux, Bluetooth or speakerphone hardware, audio services, and API credentials make setup demanding; a live call and transcript also have consequential privacy and reliability stakes.", "standout": false, "blocked": null }, { "pr": 153669, "name": "corkboard", "does": "Stores owner-only questions, choices, and tasks on a shared board, with Desktop, CLI, and slash-command answering plus kanban auto-pin and self-clearing checks.", "useful": 4, "reliable": 4, "polish": 5, "creative": 4, "beyond": 4, "total": 25, "best": "The coherent owner decision flow spans agent tools, kanban hooks, shared storage, and a notably finished Desktop interface, with broad focused tests.", "worst": "Owner-added shell checks run without an approval prompt when manually or cron-triggered; documentation clearly identifies this power, but safe operation depends on treating those commands like crontab entries.", "standout": false, "blocked": null }, { "pr": 153851, "name": "foreman", "does": "Gates coding-agent changes that weaken tests or checks, runs a deterministic diff detector and optional second-model review, and leaves merge approval to the person.", "useful": 5, "reliable": 4, "polish": 5, "creative": 5, "beyond": 5, "total": 29, "best": "The deterministic, diff-based gate targets a real failure mode in coding agents and pairs it with CI, review records, a Desktop panel, and extensive tests.", "worst": "Its open-command threat model has documented bypasses, and its effectiveness depends on protected base branches and CI; the CI workflow itself can still be changed in a pull request.", "standout": true, "blocked": null }, { "pr": 149183, "name": "clack", "does": "Adds nine Hermes tools for encrypted Clack agent-to-agent messaging, invitation onboarding, polling, acknowledgement, and reconciliation over a public or self-hosted relay.", "useful": 3, "reliable": 4, "polish": 4, "creative": 5, "beyond": 4, "total": 24, "best": "End-to-end encrypted, signed messaging with explicit trust-on-first-contact avoidance and durable acknowledgements gives agents a distinct coordination channel.", "worst": "It depends on an external relay and Linux, and initial setup requires independently verified relay and inviter keys, adding real onboarding friction.", "standout": false, "blocked": null }, { "pr": 152159, "name": "genui", "does": "Renders schema-validated, live-updating interactive cards inside Hermes Desktop and sends a user’s card decisions back into the agent turn.", "useful": 4, "reliable": 4, "polish": 5, "creative": 4, "beyond": 5, "total": 27, "best": "It changes the agent interface from chat into persistent interactive components with action callbacks, a sandboxed renderer, schema validation, and polished demos.", "worst": "It is Desktop-specific and depends on the agent consistently invoking and updating the cards; it also has a jsonschema dependency.", "standout": true, "blocked": null }, { "pr": 153144, "name": "signet", "does": "Drafts email for Hermes and releases it only after a person approves that exact message; it fingerprints the approved content and blocks common direct mailer paths.", "useful": 4, "reliable": 4, "polish": 5, "creative": 5, "beyond": 5, "total": 28, "best": "The seal is tied to the canonical message bytes, with approval and post-approval checks, credential attachment guards, audit records, and extensive tests.", "worst": "The direct-send guard only covers recognized mail tools and pathways rather than acting as a system sandbox, and actual delivery requires configuring a mail transport.", "standout": true, "blocked": null }, { "pr": 151875, "name": "patchsmith", "does": "Hands a bounded coding task to a local model, returns or applies its diff, and can run only a user-configured test command.", "useful": 4, "reliable": 4, "polish": 4, "creative": 4, "beyond": 4, "total": 24, "best": "A local model can handle mechanical edits without sending project code to a hosted model; proposal mode, path protections, bounded budgets, and secret-minimal subprocess environments support safer use.", "worst": "It requires a local tool-calling model and compatible endpoint; only Ollama is described as measured, and the entry’s effectiveness depends on the quality of that model and owner-configured checks.", "standout": false, "blocked": null } ] }