{"batch":"read1-125","plugins":[{"pr":0,"name":"done-bell","does":"Adds a native Hermes notification bell that records task completion context and exposes a small desktop/dashboard view.","useful":3,"reliable":4,"polish":4,"creative":4,"beyond":4,"total":23,"best":"It is a genuine Hermes-native completion and context surface with tests, a dashboard manifest and a desktop component.","worst":"Its value is a narrow notification workflow rather than a broadly necessary capability.","standout":false,"blocked":null},{"pr":0,"name":"homeassistant","does":"Connects Hermes to Home Assistant and bundles a broad, tested collection of local smart-home and device integrations.","useful":5,"reliable":5,"polish":5,"creative":4,"beyond":5,"total":29,"best":"It turns Hermes into an agent for a real local automation platform with substantial code and integration tests.","worst":"The reference bundle is broad and polished but largely an integration layer over an existing platform rather than a new agent concept.","standout":true,"blocked":null},{"pr":153963,"name":"hermes64","does":"Lets Hermes plan, build, validate, playtest and export small late-1990s-style 3D adventure games in Godot.","useful":4,"reliable":4,"polish":5,"creative":5,"beyond":5,"total":28,"best":"It makes Hermes itself a game-making system, with many native tools, skills, story checks, bot playtests, assets and a desktop surface.","worst":"The scope is ambitious and setup is heavy: Godot and Python dependencies are required, platform support is limited, and human playability is not proven.","standout":true,"blocked":null},{"pr":153653,"name":"ghl-command","does":"Adds twelve native GoHighLevel CRM tools for contacts, pipelines, opportunities, email, SMS and workflow enrollment.","useful":4,"reliable":3,"polish":4,"creative":2,"beyond":3,"total":19,"best":"It exposes practical CRM actions through Hermes secret storage and has explicit argument validation and error responses.","worst":"The implementation has only a small registration test despite exposing consequential delete, messaging and workflow actions, with no visible confirmation layer.","standout":false,"blocked":null},{"pr":149917,"name":"hive-intelligence","does":"Connects Hermes to a hosted Hive MCP service offering broad crypto prices, research, wallet, security, DeFi and prediction-market tools.","useful":4,"reliable":3,"polish":4,"creative":3,"beyond":4,"total":22,"best":"Its discovery loop and broad normalized crypto-tool catalog can save an agent from stitching together many providers.","worst":"The submitted package is primarily remote configuration and skills, so correctness, availability and much of the security boundary depend on a proprietary hosted service outside the pinned code.","standout":false,"blocked":null},{"pr":153954,"name":"defuzz-componentize","does":"Guides turning Python scripts into reusable components, compares saved baseline output, and applies bounded project kits with a desktop settings view.","useful":3,"reliable":4,"polish":4,"creative":4,"beyond":4,"total":23,"best":"It provides unusually concrete component contracts, evidence, path-safety checks and many tests around a code-organization workflow.","worst":"It is a specialized developer workflow with Linux/Python 3.13 requirements, and its runner can execute user-supplied commands or imports with host permissions.","standout":false,"blocked":null},{"pr":150751,"name":"restoreproof","does":"Restores backups in an isolated Docker network, boots the app, checks declared customer workflows and produces evidence-backed recovery verdicts.","useful":4,"reliable":4,"polish":5,"creative":5,"beyond":5,"total":28,"best":"It tests the claim that a backup works against real data-linked workflows, with clear PASS/PARTIAL/FAIL/BLOCKED semantics, approvals, cleanup, reports and a polished Desktop view.","worst":"A safe useful drill still requires Docker, an app image, backup inputs and a carefully declared contract, so setup is substantial and operational edge cases remain.","standout":true,"blocked":null},{"pr":150178,"name":"payops","does":"Reads Paystack transactions and settlements and produces deterministic, read-only reconciliation and data-quality reports.","useful":3,"reliable":4,"polish":3,"creative":3,"beyond":3,"total":19,"best":"It is a focused financial-operations tool with bounded pagination, secret-store use, HTTPS GETs, deterministic semantics and unit/provider tests.","worst":"The catalog release supports only Paystack and has no ledger matching, continuous monitoring, alerts or other higher-level operations, so its day-one audience is limited.","standout":false,"blocked":null},{"pr":152619,"name":"proofly","does":"Verifies claimed Base transfers against read-only chain evidence and corrects tool results or final replies when token, recipient or raw amount do not match.","useful":5,"reliable":5,"polish":5,"creative":5,"beyond":5,"total":30,"best":"It addresses a distinctive agent failure mode with exact structured intent, chain-id and log checks, conservative verdicts, replay-safe correlation, tests and output correction.","worst":"It is limited to Base mainnet and allowlisted payment-tool conventions, so coverage must be configured and expanded for other chains and transaction shapes.","standout":true,"blocked":null},{"pr":151427,"name":"agent-autocommit","does":"Automatically commits each agent turn's own hunks locally, isolates concurrent sessions, records task rationale and writes WHY.md summaries on completion.","useful":4,"reliable":4,"polish":5,"creative":5,"beyond":5,"total":28,"best":"It changes Hermes's collaboration and accountability model with direct Git plumbing, profile identity, secret-name filtering, no-push defaults and a demo plus extensive tests.","worst":"It performs implicit repository writes on lifecycle hooks and has a complex concurrency and configuration surface where unusual Git states can still be difficult to reason about.","standout":true,"blocked":null},{"pr":153577,"name":"seatbelt","does":"Interposes policy checks before trading-looking tool calls to allow, block, ask the human or clamp leverage, with paper mode, limits, logs and halt controls.","useful":5,"reliable":4,"polish":5,"creative":5,"beyond":5,"total":29,"best":"It adds a reusable safety boundary to Hermes across exchanges and trade-like tools, defaults to paper mode, fails closed on uncertainty and has strong pure-engine and integration tests.","worst":"It depends on adapters and Hermes approval-hook behavior, cannot see exchange-app actions, and its own README acknowledges MCP-hook coverage is untested.","standout":true,"blocked":null},{"pr":151660,"name":"aitrainingdatabroker","does":"Adds a read-only remote MCP blog server and one skill for questions about licensing company data for AI training.","useful":2,"reliable":3,"polish":3,"creative":2,"beyond":3,"total":16,"best":"The package is transparent, small and safe by scope: remote read-only access, no key, no local execution and links back to source articles.","worst":"It is mostly a remote blog connector plus instructional skill, so a prompt/skill and ordinary web access cover much of the same value and there is little pinned implementation to test.","standout":false,"blocked":null}]}