All plugins

skill-workshop

Get new skills for Hermes safely: mine them from your past chats (Skill Miner), build them from a description (First Principles Skill Builder), and guard against prompt injection in skills and content you bring in (Taint Shield). Scans skill writes before they land and holds back flagged external skills at first use (Skill Guard); Workshop installs need your approval. Disclosure: Skill Miner reads the Hermes session database read-only through the public hermes_state module (tool names and counts, plus the first request of each session in abstracted, redacted form; for one session you choose, also the ordered tool steps as argument shapes and your own messages as redacted quotes, to draft a skill from it); Skill Builder runs --version on a fixed list of common tools and nvidia-smi with a minimal environment and reads memory, disk space, skill names and model-cache folder names; Taint Shield watches tool calls in memory; Skill Guard reads the text of skills in your skills folders (and external skill folders named in config.yaml, read through Hermes's config loader) when they are installed or loaded and keeps a hash and a decision for each in plugin storage. Drafts, revisions and the trusted list live in plugin storage; installing a skill writes one file into the skills folder after your approval. Two install approvals, the Shield's block-or-ask rules and Skill Guard's install scan use pre_tool_call; Skill Guard also uses transform_tool_result and pre_llm_call (hold back a flagged skill, once) and on_skill_lifecycle (scan a skill after Hermes reports it installed). No network calls.

flabanabba8Toolsv0.4.0Updated 5 days ago

About

Get new skills for Hermes safely: mine them from your past chats (Skill Miner), build them from a description (First Principles Skill Builder), and guard against prompt injection in skills and content you bring in (Taint Shield). Scans skill writes before they land and holds back flagged external skills at first use (Skill Guard); Workshop installs need your approval. Disclosure: Skill Miner reads the Hermes session database read-only through the public hermes_state module (tool names and counts, plus the first request of each session in abstracted, redacted form; for one session you choose, also the ordered tool steps as argument shapes and your own messages as redacted quotes, to draft a skill from it); Skill Builder runs --version on a fixed list of common tools and nvidia-smi with a minimal environment and reads memory, disk space, skill names and model-cache folder names; Taint Shield watches tool calls in memory; Skill Guard reads the text of skills in your skills folders (and external skill folders named in config.yaml, read through Hermes's config loader) when they are installed or loaded and keeps a hash and a decision for each in plugin storage. Drafts, revisions and the trusted list live in plugin storage; installing a skill writes one file into the skills folder after your approval. Two install approvals, the Shield's block-or-ask rules and Skill Guard's install scan use pre_tool_call; Skill Guard also uses transform_tool_result and pre_llm_call (hold back a flagged skill, once) and on_skill_lifecycle (scan a skill after Hermes reports it installed). No network calls.

Skill Workshop

Skill Workshop

Get new skills for Hermes safely. Skill Workshop scans every skill before it is installed or used, including ones from other people, and it helps you mine skills from your past chats and build them from a description.

New in 0.4.0: Skill Miner can turn one session into a skill. Say "turn what we just did into a skill": miner_session returns the ordered steps that ran (argument shapes only, retries and polling collapsed, failures with what replaced them) and your own corrections quoted with their message numbers, split into standing preferences and one-off facts by a plain word-pattern check that can be wrong. A draft with a separate "Preferences from " section goes through the usual refine, check, scan and confirm path.

New in 0.3.2: Shield recognizes commands in Markdown image labels, JSON string values and shell continuations; exact trusted URLs now work the trust confirmation names the destination, and approved skills keep their decision when a linked file is viewed.

New in 0.3.1: Shield recognizes command-shaped instructions embedded in HTML and prose; Skill Guard names the skill and decision in approval prompts and identifies batched skill writes.

New in 0.3.0: Skill Workshop scans every skill before it is installed or used, including ones from other people. When Hermes writes a skill, or a file in your skills folder, the new text is scanned first and you see each finding (file and line) in the approval box. When a skill is loaded, it is scanned again: a skill with hidden text, instructions aimed at your agent, or commands that send secrets out is held back from the agent, and the agent tells you and waits for your decision (approve this exact version, or quarantine it). A skill added any other way (hermes skills install, copying a folder in) is scanned as soon as Hermes reports it, and at its first use. Ask "scan my installed skills" to see every skill as clean or flagged, with file and line. Your decisions are remembered by a hash of the skill's files, so a clean or approved skill is not asked about again, and any change to a file is scanned again. Skill Miner now names a mined skill from what you asked, and ignores the housekeeping calls (finding a skill, searching files) that open almost every chat.

At a glance

  • Who it is for: people who want their Hermes agent to learn new skills, and who worry about what an imported skill or a web page could make the agent do.
  • Install: hermes plugins install flabanabba8/skill-workshop, then hermes plugins enable skill-workshop, then restart Hermes.
  • Needs: Hermes 0.22 or later. No API keys, no extra tools, no Python packages. Skill Miner needs some past Hermes sessions to look at.
  • Internet: nothing. None of the four parts makes a network call or a model call of its own.
  • Platform: Linux is the only platform checked.
  • Safety: every skill is scanned for prompt injection before it is installed or used (a plain pattern check, not proof). Nothing is installed as a skill without a person confirming, a flagged skill is not handed to the agent until you decide, and the Shield and the Guard only block or ask; they never bypass a Hermes approval.

The four parts

Skill Miner and Skill Builder are two ways to get a skill. Taint Shield sits with them because prompt injection is a real fear with imported skills and untrusted content: a skill, web page or file can hide instructions meant for your agent, not for you.

PartWhat it is forTools
Skill MinerLooks through your recent Hermes sessions for sequences of tool calls you repeat, checks each on sessions held out of the mining, and drafts a skill for you to review. The host agent then diagnoses, refines and re-checks the draft.miner_scan, miner_session, miner_propose, miner_diagnose, miner_refine, miner_check, miner_install
First Principles Skill BuilderHelps the agent write a skill from a description so it fits the tools, hardware and permissions you actually have: it discovers resources, lints the draft, records three contrasting-case walk-throughs and keeps a revision history.skillbuilder_resources, skillbuilder_save, skillbuilder_check, skillbuilder_case, skillbuilder_install
Skill GuardScans every skill before it is installed or used, including ones from other people. It checks what Hermes is about to write into the skills folder, checks each skill again when it is loaded (a flagged, unapproved one is replaced by a plain warning naming the findings), scans skills added by other routes when Hermes reports them, lists every installed skill as clean or flagged, and remembers your decisions by content hash.skillguard_scan_installed, skillguard_decide
Taint ShieldA prompt-injection guard. It remembers which URLs, addresses and command lines came from untrusted web, browser, mail or file content, and blocks or asks before the agent uses them in a shell command, a file write or an outgoing message.shield_report, shield_scan, shield_trust

That is 17 tools, and five hooks: pre_llm_call, post_tool_call, transform_tool_result, on_skill_lifecycle, and pre_tool_call (registered four times: the Shield's check, the Guard's install scan, and the install approvals of Skill Miner and Skill Builder). Each part also registers its skill (skill-miner, first-principles-skill-builder, taint-shield, skill-guard). Nothing else is registered: no commands, no middleware, no desktop or dashboard surface.

Each part behaves exactly as its stand-alone plugin did; this repository bundles the three into one plugin so they can be installed together (Skill Guard is new in 0.3.0). A short example: ask Hermes "find a workflow I repeat and turn it into a skill" (Skill Miner), or "build me a skill that turns meeting recordings into notes on this machine" (Skill Builder). While either runs, the Shield watches what the agent reads and does.

What each part does

Skill Miner

You solve the same kinds of tasks again and again. Skill Miner reads Hermes's session database (opened read-only) for tool-call sequences that recur, checks that each holds up, and drafts a skill. It proposes only.

  • miner_scan reads tool names and counts, plus the first user request of each session in abstracted and redacted form (paths, URLs, long ids and numbers replaced; secrets redacted by pattern). That text is used only to find wording that sessions share and comes back as request_terms. Sequences of 3 to 6 steps that appear in at least min_support sessions (default 3) are candidates (a two-step chore that writes a file is accepted when your requests share wording); housekeeping calls that open almost every chat (listing or viewing skills, searching tools or files) are left out, and candidates are ranked by what is specific to the job: your shared request words, the file types touched and the write steps. A drafted skill is named and described from your request; mining uses two thirds of your sessions and the rest are held out. A recommend verdict needs enough support, at least 75 percent consistency, and at least one held-out hit.
  • miner_session mines ONE session instead of many (the current or most recent by default; session picks an id, a part of a title or a date; list shows recent sessions). It returns the tool steps that ran in order with argument shapes only (kinds of value, never the values), retries and repeated polling collapsed to one step, failed steps marked with the step that replaced them, and your messages split into standing preferences (always, never, make sure, I want, instruction verbs and corrections after a tool result) and one-off facts such as a file or URL. Each item is a quote with its message number and the reason it was classified; paths, URLs, emails and secrets are replaced. The split is a plain heuristic and can be wrong, so the person decides, and quotes and tool results that look like instructions are flagged as untrusted and kept out of the draft. It saves a draft with a "Preferences from " section as an ordinary proposal; it makes no model call and installs nothing. A routine done once has no held-out check (verdict single_session).
  • miner_propose drafts a SKILL.md for one pattern, with abstracted typical arguments (paths, URLs, long ids and numbers replaced, secrets redacted by pattern) and up to three of your original request texts. This step reads the original request text of the supporting sessions (redacted).
  • miner_diagnose returns bounded (8000 characters), redacted context for one proposal or revision for the host agent to reason over: structural problems in the draft and where its supporting sessions hit errors or retries.
  • miner_refine saves the agent's revised text as a numbered revision beside the original, which is never modified.
  • miner_check runs deterministic checks on a revision and replays its step order on held-out sessions. A weak mining verdict stays weak.
  • miner_install writes the proposal into your skills folder as <name>/SKILL.md. A revision must have passed miner_check unchanged; it refuses to overwrite an existing skill and always asks a person to confirm.

Limits: it mines tool-name sequences, not intent, so a pattern is a habit and not necessarily a good skill; replay compares step order only; held-out checking is one deterministic split, not statistics; redaction is pattern based and can miss unusual secrets, so review a draft before sharing it. On the author's own 90-day history it found six candidate patterns and recommended none. That is the intended behaviour, not a failure.

First Principles Skill Builder

You ask for a skill, or you have one that only works on one machine. The method is a registered skill: establish the goal and acceptance checks, discover resources before recommending, write the workflow with discovery, setup, authorization, acceptance and a degraded mode for each capability, then walk it through three contrasting cases (limited CPU and offline; capable GPU with cached models; a hosted service you explicitly chose). The reasoning is done by the host agent; the plugin adds five deterministic helpers.

  • skillbuilder_resources: CPU count, RAM, free disk, NVIDIA GPU vendor and memory (via nvidia-smi), versions of a fixed list of common executables, cached model names, and the host's registered tools and skills. Names only: no model is loaded and no credential or account store is read.
  • skillbuilder_save: save a draft as a new numbered revision. Earlier revisions are never overwritten.
  • skillbuilder_check: one result per check, with offending lines (frontmatter, description quality, size, personal paths, mandatory provider ids, acceptance, degraded mode, retired workflows, harness neutrality, side effects without approval, referenced files and tools, recorded contrasting cases).
  • skillbuilder_case: record the agent's walk-through of one contrasting case with a verdict (works, degraded, blocked).
  • skillbuilder_install: install a clean, fully walked-through revision into your skills folder. It refuses failing checks, missing cases and overwrites, and always asks a person to confirm.

Limits: the checks are structural lint, not a judgment that a skill works; the walk-throughs are reasoning, not execution; GPU detection covers NVIDIA only.

Skill Guard

Skills from other people can hide instructions meant for your agent. Skill Guard scans every skill, whoever wrote it, with the same plain pattern check as the Shield, line by line.

  • Before it is installed: when Hermes writes a skill (skill_manage) or a file inside your skills folders, the new text is scanned first. A clean write goes through silently; a write with findings asks you, with every finding (file, line, what it is) in the approval text. Skills the workshop installs itself after a clean scan are recorded as clean.
  • When it is used: when a skill is loaded, its files are scanned. A flagged version nobody has approved is not handed to the agent: it gets a plain warning naming the findings and is told not to follow those lines until you decide, and to tell you. You are told once per version. A /skill-name command is checked the same way and is stopped before the model sees the text.
  • Added another way: hermes skills install and copying a folder in have no hook before they land, so Hermes reports the new skill afterwards and the Guard scans it then (and tells you at your next message); the use-time check holds it back either way.
  • Scan what you have: skillguard_scan_installed ("scan my installed skills") lists every skill in your skills folders as clean, approved, quarantined or flagged, with file and line for each finding.
  • Your decisions: skillguard_decide records approve (use it as it is) or quarantine (never hand it to the agent) for the exact current version, and always asks a person to confirm. Decisions are kept by content hash, so a clean or approved skill is never re-scanned or re-asked, and a changed file is scanned again.
  • Limits: a pattern check, not proof; it cannot see a malicious skill that is rephrased, encoded or hidden in an image. Files the agent writes into the skills folder with a shell command are caught at their first use, not before.

Taint Shield

Text the agent reads from the web, a browser, mail or similar sources is untrusted, and attackers hide instructions in it. Taint Shield remembers what came from untrusted content and stops the agent acting on it.

  1. URLs, domains and addresses in your own messages are trusted for that session.
  2. Output from tools whose names look like web, browser, fetch, mail, search, scrape, download and similar is untrusted; its URLs, domains, emails, IPs and command lines become tainted. Output with instruction-like phrases gets a one-line warning appended.
  3. Before a sensitive tool runs (terminal, execute_code, send/post/email tools, write_file, patch, browser navigation or typing) its arguments are checked. It blocks reading local secrets and sending them out, editing login or startup files after untrusted content was read, running a command line copied from untrusted output, and using a tainted destination after an injection-like phrase. It asks a person before using a destination that appeared only in untrusted output. Everything else is untouched.
  4. shield_trust adds a permanent trusted destination and always requires a person's approval, so injected text cannot trust itself. shield_report shows sources, flags and decisions per session; shield_scan analyses any text without changing anything, and with path it checks a skill file or folder you downloaded.
  5. Scan before install. miner_install and skillbuilder_install scan the full draft first, using the Shield's own detection, line by line. A clean skill installs and the result says scanned for prompt injection: no findings. If anything is found (injection-like instructions, hidden zero-width text or instructions in an HTML comment, a command that sends or copies a secret or changes a login file, a download piped into a shell) nothing is installed and the result lists each finding with its line and how to go on: fix the draft, or repeat the install with allow_findings set, which still asks a person to approve and shows the findings in the approval text. Ordinary shell steps in a skill's own instructions are not flagged. The scan is on the same heuristic footing as the rest of the Shield: it reads only the draft's SKILL.md (the only file these two tools install), and a rephrased or encoded injection can slip past it.

Limits: this is heuristic provenance, not proof. It matches literal values, so an injection that makes the model rephrase or encode a destination can slip past; commands you also wrote are trusted; it only watches tools it recognises by name. It reduces risk, it does not make an agent safe to run on hostile content. It adds vetoes and approval requests only and bypasses no approval. Block rules are tuned on a small hand-written corpus, not a benchmark.

Where the data goes

All four parts keep their data under one Hermes data folder, plugin-data/skill-workshop/, one sub-folder each: skill-miner/proposals/, first-principles-skill-builder/drafts/, taint-shield/trusted.json and skill-guard/decisions.json (a content hash, a skill name and a decision for each skill version; no skill text). Data written by the stand-alone plugins (under plugin-data/skill-miner/ and so on) is not read or migrated. Installing a skill writes one new <HERMES_HOME>/skills/<name>/SKILL.md, and only after you confirm.

Limits of the suite

  • Linux is the only platform checked.
  • The desktop and dashboard surfaces are not part of this plugin; none of the parts has one.
  • Taint Shield guards the tool calls of the whole Hermes session, not only the other two parts. It can ask or block on ordinary work that follows untrusted content; trust a destination with shield_trust if you want it to stop asking.
  • Live evidence for each part is the part's own (see VERIFICATION.md). The suite itself was tested offline plus the Hermes admission check.

Credits and licence

No third-party code or data is bundled: all four parts are original code with no vendored libraries, no downloaded tools and no media. The plugin does not use TMDB, ani-cli, Luffy, Live2D or any other upstream tool. Licence: GPL-3.0-or-later (see LICENSE).

AI assistance: written with AI coding assistants (Anthropic Claude and OpenAI Codex agents) under the maintainer's direction; what was actually tested is listed in VERIFICATION.md.

README from the reviewed commit 6568d0f; it updates when the author re-pins.

More by flabanabba8

6 plugins by flabanabba8 in the catalog.

flabanabba8live2d-riggerCreate a provisional Live2D character from a brief with complete image parts, local cutout, placement, an experimental Umamo/JDK rig and Core motion review; also inspect and adjust a copy of an existing model. XY head turns, independent limbs and polished eyes are not provided. Disclosure — runs the Java (JDK) and Kotlin/Umamo classpath, the Python with Playwright and the Cubism SDK viewer folder that you set in the plugin settings (the model cannot name a program, classpath or command), plus FFmpeg and a local headless browser against a viewer served only on a temporary loopback port, each with a minimal environment; on first use it scans PATH and the hardware (read-only) and asks which routes to use; opening an existing model copies its folder into the plugin's data directory; on request it downloads official Live2D sample archives from cubism.live2d.com after you agree to Live2D's licence; the host image tool may use its configured provider; no Cubism SDK, Core, viewer, sample model, layers or textures are bundled, and the catalog thumbnail contains character artwork.Communitytools1flabanabba8case-trailTrace case-history and treatment research to actual opinion passages; analyze supplied sources without a required repository. Disclosure — Sends case search queries and Justia page URLs through the web search/extract provider you configured in Hermes, and can fetch one public OpenJurist case page you select after a robots.txt check (OpenJurist sees the URL and your IP). Reads local opinion files at paths given in chat, refusing kernel and device files, the Hermes home and key stores; the plugin setting opinion_folders can limit it to folders you list. Stores fetched text, review receipts and rendered pages privately in its plugin data folder. Novelty research tool, not legal advice; unaffiliated with Justia or OpenJurist.Communitywebflabanabba8creator-studioMake media with Hermes from one request: researched lesson videos, two-host podcast episodes, captions for any recording, narrated step-by-step tutorials, and clips cut from a YouTube video or stream by reading its transcript. Bundles five plugins (YouTube Studio, Podcast Studio, Caption Studio, Tutorial Creator, Clip Studio) into one entry. Disclosure — Research pages are fetched through the Camoufox browser you set in the plugin settings, in a fresh anonymous profile (public addresses only). On first use it lists the voices and speech recognizers this computer has and asks you which to use; it never picks one for you. If you choose Gemini narration, each spoken line is sent to Google, either to Vertex AI (aiplatform.googleapis.com) with a Google sign-in that Hermes holds for your own OAuth client (hermes auth add creator-studio:google; the plugin never sees the token), or to generativelanguage.googleapis.com with the Gemini API key in this plugin's settings (GEMINI_API_KEY, the same variable Hermes' own Gemini provider uses), behind a spend guard (default 0.50 USD). To find speech servers it sends read-only GET requests (/v1/models, /openapi.json) to 127.0.0.1 only, on ports 2022, 5000, 5002, 7860, 8000, 8001, 8020, 8080, 8880 and 9000, plus any loopback address you set. Caption Studio runs the recognizer command you set in the plugin settings, or uploads audio to the HTTPS speech service you set (public addresses only) or to Groq (api.groq.com) with your key, only when you allow uploads; the model cannot set commands, endpoints, keys or upload consent. Programs it starts get a minimal environment (paths, locale, display/audio session and proxy settings), never your provider keys. Clip Studio runs the yt-dlp you installed (or set in the plugin settings) and makes requests to YouTube only when you give it a link: the video page and one caption track, then the audio track if you ask for speech recognition, the sections you choose to cut, and, if you ask what chat reacted to, the public chat replay of a finished stream (other people's messages and display names; kept in private plugin storage, the raw download deleted once counted, names kept only for paid messages). It never uses browser cookies or a YouTube sign-in. Recognizing a whole recording continues in a background thread inside Hermes until its pieces are done. Clips, joined videos, titles, descriptions and thumbnails are written to plugin storage. Runs FFmpeg and ffprobe, an installed speech program if you choose one (espeak-ng, flite, piper), and for screen tutorials xwininfo, xprop, xdotool, pactl and xclip; the recorder starts one detached FFmpeg worker per take, and tutorial steps click and type through Hermes computer_use, so its approvals apply. Stores projects privately in plugin storage. Nothing is published.Communitytoolsflabanabba8game-labGame Lab combines ModLab, Game Localization Studio, Sprites and experimental Playtest Pilot. ModLab works on a private game copy and invokes separately installed tools; localization drafts and exports RPG Maker MV/MZ text and the text of Unity Mono builds (JSON text assets, script-component strings and Assembly-CSharp string literals; patched copies of changed files only, the game folder is never written); localization companions estimate width and row budgets, merge and audit translation data, and proofread English; Sprites inspects and previews character sheets. Playtest Pilot uses approved X11 input, captures screen pixels within the selected window, stores private evidence frames and returns reduced previews and optional local OCR hints; for a failed check it can build a short captioned MP4 from those frames with your FFmpeg, or cut one from a recording file you point to (that asks for approval). On first use ModLab reads Steam's library list and app manifests to find installed games, lists the files in the chosen game folder, runs read-only hardware checks, and asks which modding route to use. It never fills a disk: a free-space reserve, a total cap, pruning of old revisions and a confirm-first cleanup tool. The Universal Modder and REA programs are the ones you set in the plugin settings or find on PATH (the model cannot name a program), and each toolkit call and each playtest check asks for your approval. Every program it starts gets a minimal environment, never your provider keys. The plugin makes no network calls of its own. Disclosure — Playtest Pilot sends synthetic key presses and clicks to the X11 window you select (via xdotool and Hermes computer_use, each check or batch approved), captures that window's pixels and stores evidence frames in the plugin's data folder; ModLab reads your Steam library and app-manifest files to list installed games, copies the game folder you choose into plugin data (with a disk cap), Game Localization Studio reads the RPG Maker data file or Unity game folder you point it at and copies the files that hold text into plugin data, writes a JSONL units file into the draft folder, and reads translations back from the JSONL file path named in the tool call (absolute path; symlinks, credential folders and the Hermes home are refused); and, after your approval of each call, runs the Universal Modder / REA programs set in the plugin settings or found on PATH.Communitytools

Developer Mode

Explore built-in Hermes toolsets and actions.

HermesAgent

Developer?

Create a Hermes extension for Desktop or agent workflows.

Developer Section