{ "batch": "read1-029", "plugins": [ { "pr": 0, "name": "done-bell", "does": "Rings an OS notification/chime and shows a busy/done/needs-you chip when a Hermes turn finishes or waits for approval.", "useful": 3, "reliable": 4, "polish": 4, "creative": 2, "beyond": 3, "total": 19, "best": "Small, keyless Desktop+CLI quality-of-life plugin with hooks, a pane, slash commands and tests.", "worst": "Simple idea; README install steps are Windows-dev paths and require manually copying the Desktop pane.", "standout": false, "blocked": null }, { "pr": 0, "name": "homeassistant", "does": "Official Home Assistant gateway platform plus four REST tools to read and control smart-home devices.", "useful": 4, "reliable": 5, "polish": 4, "creative": 3, "beyond": 5, "total": 26, "best": "Mature gateway adapter + tools with a broad test suite and secret-store env with password prompt.", "worst": "Port of a formerly built-in feature rather than something new.", "standout": false, "blocked": null }, { "pr": 151510, "name": "marsell", "does": "Agent Plugins package pointing Hermes at Marsell's hosted marketing/sales MCP server.", "useful": 2, "reliable": 2, "polish": 3, "creative": 1, "beyond": 2, "total": 12, "best": "Marked untrusted with OAuth so Hermes asks before write calls.", "worst": "No plugin code at all: a generic MCP connection manifest for a private SaaS backend, repo is cross-client registry metadata.", "standout": false, "blocked": null }, { "pr": 151027, "name": "tripwire", "does": "Hooks that append refused tool calls and approval denials to a private local events file with minimal metadata.", "useful": 2, "reliable": 3, "polish": 1, "creative": 2, "beyond": 3, "total": 14, "best": "Careful privacy-minimal logging (O_NOFOLLOW, 0600, no args stored).", "worst": "No README and no tests in the plugin folder; ~140 lines with narrow value.", "standout": false, "blocked": null }, { "pr": 152497, "name": "hypruse", "does": "Computer-use provider that replaces Hermes's driver on Hyprland with the hypruse CLI using accessibility trees and agent-owned workspaces.", "useful": 3, "reliable": 4, "polish": 4, "creative": 4, "beyond": 5, "total": 25, "best": "Fixes a real Hermes gap (cua-driver failing on Wayland/Hyprland) by swapping the computer_use backend, backed by an extensively tested CLI.", "worst": "Niche audience (Hyprland users only) and requires a separate hypruse install.", "standout": true, "blocked": null }, { "pr": 151541, "name": "skillens", "does": "Watches tool calls against the numbered steps of loaded skills, shows a fidelity checklist pane in Desktop and nudges the agent about skipped steps.", "useful": 4, "reliable": 3, "polish": 4, "creative": 5, "beyond": 4, "total": 24, "best": "Novel, cheap (no LLM judge) check that the agent actually follows its skills, with Desktop pane and screenshots.", "worst": "No tests found; heuristic step matching may miss or misjudge steps.", "standout": true, "blocked": null }, { "pr": 151263, "name": "timetopost", "does": "Agent Plugins package with skills and an MCP config for TimeToPost's hosted social-scheduling server.", "useful": 2, "reliable": 3, "polish": 3, "creative": 1, "beyond": 2, "total": 13, "best": "Honest disclosure, OAuth through Hermes, server marked untrusted with human approval for writes.", "worst": "Essentially skills plus a hosted-service MCP URL; the tool list is not in the pinned code.", "standout": false, "blocked": null }, { "pr": 153605, "name": "treasury-cash-forecast", "does": "Builds a 12-month direct cash-flow forecast workbook from a template using ETS/SARIMA models, with scenarios and cent-level reconciliation.", "useful": 2, "reliable": 5, "polish": 4, "creative": 3, "beyond": 3, "total": 20, "best": "Deterministic, tested forecasting engine (67 tests) so the model never computes numbers; offline, no credentials.", "worst": "Narrow corporate-treasury audience.", "standout": false, "blocked": null }, { "pr": 150106, "name": "load-lock", "does": "Machine-wide counting semaphore with 17+ tools, heartbeats and Desktop/dashboard UI so parallel coding agents take turns on heavy builds.", "useful": 3, "reliable": 3, "polish": 5, "creative": 4, "beyond": 4, "total": 23, "best": "Polished cross-agent coordination with dashboard, Desktop pane, screenshots and session hooks.", "worst": "Relies on an external daemon outside the plugin folder; no tests in the plugin folder and a heavy tool surface.", "standout": false, "blocked": null }, { "pr": 152263, "name": "besthumanizer", "does": "Tools that score and rewrite text for AI-likeness via besthumanizer.app's MCP API, plus a hook on outgoing post/send tool results.", "useful": 2, "reliable": 3, "polish": 3, "creative": 2, "beyond": 2, "total": 14, "best": "Deterministic before/after scoring keeps the rewrite checkable.", "worst": "Thin client for a metered paid service; no tests, and the transform_tool_result hook touches every post-like tool result.", "standout": false, "blocked": null }, { "pr": 149954, "name": "tensorfold", "does": "Registers TensorFold's local inference server as a Hermes model provider and adds a dashboard stats tab.", "useful": 2, "reliable": 3, "polish": 3, "creative": 3, "beyond": 4, "total": 19, "best": "Brings a local inference engine into Hermes's model picker on the user's own hardware.", "worst": "Very small (~46-line provider), no tests, tied to one niche engine; README has users echo the key into ~/.hermes/.env.", "standout": false, "blocked": null }, { "pr": 150029, "name": "airlock-sandbox", "does": "Terminal backend routing Hermes commands through an external Airlock policy engine and Docker sandbox with Ed25519-signed receipts.", "useful": 3, "reliable": 4, "polish": 4, "creative": 4, "beyond": 5, "total": 25, "best": "Fail-closed design with 41 tests, signed verifiable receipts and a demo video, via the terminal provider seam.", "worst": "Depends on running the author's own Airlock service; few users will.", "standout": false, "blocked": null } ] }