Recently Added
Fresh skills just added to the collection. Be the first to try them out!
omni-to-hodos-cutover
by toxicwind
Procedure for cutting over tau agent configuration from obsolete OmniRoute to VansRouter/hodos gateway system
fix-gatehouse-mcp-auth
by toxicwind
Fix MCP gatehouse authentication by setting require_mcp_auth to false in config file
estate-submodule-update
by toxicwind
Procedure for adding a submodule to estate repository and updating README with infrastructure changes
issuelooper-worker
by kazunori279
"The worker of this repository's issue loop (issuelooper), for a session whose CC_ROLE is already worker: one claimed issue from trace to pull request. Worktree from origin/main (or from the branch an earlier worker left), cause traced to a line, a failing pytest first (Fail-before), the fix, the sweep, the PR in the loop's format, stage:pr. Also covers updating a PR that came back, and releasing a claim. Do not load this to find out what your role is."
issuelooper-tester
by kazunori279
"The tester of this repository's issue loop (issuelooper), for a session whose CC_ROLE is already tester. Two units: the live check of one stage:live issue against the live environment, and the UX run, which walks the deployed app the way a person would (desktop and phone, every language the app offers, screenshots while work is in progress, scrolling, reload), reads every screenshot together against the spec as a harsh UX reviewer, and files at most five new findings; and, when the workers have nothing to do, the spec check, which tests the app against the spec one requirement at a time. Do not load this to find out what your role is."
issuelooper-monitor
by kazunori279
"The monitor of this repository's issue loop (issuelooper), for the session whose CC_ROLE is already monitor: the moderator of the whole backlog. Triages every new issue so a worker can claim it, keeps priorities right around the minimum viable product, holds or closes low-value work with a reason, keeps dependencies right, closes stage:live issues once the tester verified them, decides pipeline problems itself, and raises the owner's product decisions at stage:owner. Do not load this to find out what your role is."
issuelooper-merger
by kazunori279
"The merger of this repository's issue loop (issuelooper), for a session whose CC_ROLE is already merger: resolve one pull request that conflicts with main. Rebase in a worktree, resolve so both sides survive, re-sweep, run the review gate, push, and leave the merge to the supervisor. Hand back only when the rebase needs a decision about behavior. Do not load this to find out what your role is."
openagents-api
by OpenAgentsInc
Call the OpenAgents API (Open Responses and Chat Completions, many models, one key) and read its docs. Use when a task needs a model through OpenAgents, or asks what OpenAgents models cost.
nostr
by OpenAgentsInc
How Nostr works in this repository: the three NIP lanes under nips/, the wire protocol (NIP-01), relay authentication (NIP-42), payload encryption (NIP-44), the Block agent NIPs the relay serves, and the OpenAgents NIPs (NIP-CJ jobs, NIP-CAP capabilities, NIP-PRG programs) that carry a Coder turn to a worker. Read it before touching crates/nostr, crates/nostr-relay, crates/coder-runtime/src/relay.rs, coder-worker, or any file under nips/, and before debugging a relay handoff.
google-developer-style
by OpenAgentsInc
Apply the Google Developer Documentation Style Guide to user-facing docs, READMEs, AGENTS.md, and commit messages. Use when writing or reviewing prose in this repo.
decision-api
by OpenAgentsInc
Call the OpenAgents decision API: authenticate with an oak_ bearer key, post state + typed questions to POST /v1/systemone, read Noul, Choice, and Score answers with their probabilities, and handle typed refusals, Retry-After, and idempotent retries. Use it when an agent or caller needs a programmatic decision — routing, tagging, ranking, verification — from a deployed door, or when writing client code against crates/jev or the oak CLI.
better
by OpenAgentsInc
Push the current task, or a prompt the user gives, to a deeper version of itself. Use only when the user explicitly types /better or $better.
markitdown
by remmmi
Aiguillage pour TOUT document non-markdown (PDF, Word .docx, Excel .xlsx/.xls, PowerPoint .pptx, Outlook .msg, HTML, CSV, JSON, XML, EPUB, ZIP). Appliquer ce skill dès qu'un tel fichier entre en jeu, QUEL QUE SOIT le besoin : il tranche et branche tout seul. Cas SOURCE DE CONTENU (lire, chercher, citer, résumer le fond, ou produire quelque chose à partir du contenu : rédiger un article à partir d'un rapport PDF, écrire une fonction conforme à une norme fournie en PDF, répondre à une question dont la réponse est dans un .docx ou un .xlsx) : conversion en Markdown par markitdown (Microsoft), puis lecture du Markdown. Cas OPÉRATION SUR LE FICHIER (éditer, compresser, convertir vers un autre format, expertiser : métadonnées, structure, signatures, générateur qui a produit le fichier) : le skill s'arrête et rend la main aux outils dédiés. Il s'annonce en une ligne dans les deux cas. À utiliser même si l'utilisateur ne nomme pas markitdown et dit juste « ce document », « ce PDF », « ce tableau Excel », « cette présentation ».
herdr-vues
by remmmi
Ouvrir à côté de la session, dans un pane herdr (ou tmux), ce que l'utilisateur doit SAISIR, LIRE ou VOIR VIVRE, au lieu de le mettre dans le chat. Invoquer SANS attendre qu'on le demande dès que : il faut un secret (mot de passe, jeton, clé, htpasswd) → jamais dans le chat ; l'utilisateur doit coller un résultat venu d'ailleurs (relevé d'un autre outil, export) ; un texte de plus de ~15 lignes est à relire ou corriger (brouillon de mail, d'issue, de PR, réponse à envoyer, doc, texte et son décodage) ; une commande longue ou un log à suivre (deploy, import, tests, tail -f, watch, sous-agent). Aussi sur « montre-moi en entier », « affiche », « ouvre dans nano », « ouvre en pane », « pane token », « je veux voir », « surveille », « ferme les panes ». Outil : commande vue.
simple
by remmmi
Rejouer la dernière explication dans un registre concret quand l'utilisateur ne l'a pas comprise. Déclenche sur /simple:simple, et sur toute formulation d'incompréhension : "je comprends pas", "j'ai rien compris", "c'est quoi X", "réexplique", "de quoi tu parles", "et donc ?", "en clair", "tu parles de quoi", "je te suis pas", "redis-moi". N'ajoute JAMAIS de détail : change de registre.
integration-maquettes
by remmmi
Vérifier qu'une intégration (site, thème, page, écran) correspond à sa maquette, section par section, par mesure et superposition, puis corriger ce qui compte. Déclenche sur « compare le site à la maquette », « écarts maquette / intégration », « c'est fidèle à la maquette ? », « audit visuel », « pixel perfect », et en fin d'intégration d'une maquette (phase finale de newWP). Pas pour concevoir une maquette, ni pour une non-régression entre deux versions du même site.
decode
by remmmi
Faire comprendre à l'utilisateur, paragraphe par paragraphe, un texte technique écrit par quelqu'un d'autre : réponse de mainteneur, issue, PR, changelog, doc, message d'un autre agent ou d'un LLM. Déclenche sur /decode:decode, et sur "traduis-moi ça", "décode", "je comprends pas ce qu'il veut", "qu'est-ce qu'il dit", "explique-moi sa réponse point par point". Ne résume JAMAIS : chaque paragraphe est rendu.
chatroom
by remmmi
Rejoindre un salon Agent Room quand l'utilisateur colle une URL agent-room.com/j/... ou /r/..., un code XXX-XXX-XXX, ou dit "rejoins la salle". Entrer focus, terse, pret a executer, et n'en sortir que sur "au revoir" de l'hote. Pas sur une URL seulement montree pour relecture.
claude-bridge
by remmmi
Use when a Claude on one machine must talk to a Claude on another machine but there is no inbound NAT or open port to reach it (laptop behind NAT, VPS reachable). Establishes a bilateral text channel through an SSH file mailbox on a reachable VPS. Triggers: "faire dialoguer les deux claude", "canal entre le claude du vps et moi", "parler a l'autre claude", cross-machine agent handoff.
skill-home-and-provenance
by musterd-io
Where a reviewed skill lives when more than one agent harness has to read it, and what you owe an upstream you adapted. Use when adding a skill to a repo, when the same guidance has started to exist in two places, or when vendoring skill material from someone else's repository.
seat-workspace-identity
by musterd-io
Give every agent its own workspace with its own git identity, so "who wrote this commit" is answerable without asking anyone — and know the exact line where that goes wrong, because repo-local git config is shared across work trees and the last one written owns everybody's commits. Ships a checker that tells a scoped identity from a shared one. Use when running more than one agent on a repo, when commits carry the wrong author, or when setting up agent workspaces.
pre-registered-watch
by musterd-io
A measurement that takes days is a pre-registered question with an owner and a death date — not a recurring sweep nobody reads. States the falsifier, the population, and what voids its own window before collection starts; cannot be renewed in place; and "nobody looked" is a recorded outcome. Use when a claim is about a rate or a recurrence, when installing a cron job or a dashboard, or when a decision rests on a snapshot of something that moves.
platform
by musterd-io
The platform role — the member who touches running infrastructure (service lifecycle, restarts, refreshes, deploys, shared checkouts, migrations), supervises the automated actors that touch it, and is who the team routes infrastructure trouble to. Use for diagnosing an outage or a slow or wedged service, restarting or refreshing a shared service, deploying, running or reviewing a migration, supervising automation, designing health checks, or verifying a fix is live.
observer
by musterd-io
The observer role — a read-only member who watches team state (roster, board, goals, reports, the event stream) and turns it into digests and dashboards for a named reader. It holds no lanes and sends no messages. Use when you hold the observer role, or when asked to summarize how a team's work is going without taking part in it.