Changelog

New updates and improvements to the OPVS platform.

workspace-skills-v0.4.3Nov 20, 2026

@spideriq/workspace-skills v0.4.3 — Workspace admin + film

I shipped @spideriq/workspace-skills v0.4.3 on OPVS so you can watch the 57.8-second chalkboard walkthrough and inspect the complete visual architecture right on the marketplace card before handing SpiderIQ account chores to your agent. Instead of clicking through settings tabs every Monday or paying operations help to babysit seats and keys, your agent runs 23 client-PAT workspace actions straight from chat across the brands you belong to. What's new in v0.4.3 Visual marketplace card kit. I added the Monday onboarding chalkboard hero and three architecture panels covering listBrands role verification, masked provider-key testing, read-and-redirect billing, and multi-brand setup. Narrated 57.8-second walkthrough. Carried forward from v0.4.2, the embedded minimal player shows one chat prompt creating a client brand, inviting a hire, off-boarding a contractor, and testing an API key over coffee. Vault-aligned key integrations. Building on v0.4.0 and v0.4.1, createIntegration sends provider_name, credentials, key_label, request-count daily_limit, and billing_mode with listSubscriptionTiers while keeping subscription keys private to each brand. Clerk, WorkOS, and Stripe guardrails. Every mutation checks integer brand_id and owner or brand_admin membership first, 7-day invitations stay send-then-accept, and billing redirects to /dashboard/plans with 429 versus 402 spend diagnostics. Trigger it from chat with "set up my new client brand" or "invite a teammate to my workspace". Live on the marketplace as @spideriq/workspace-skills.

OPVSRead →
mail-skills-v0.17.3Nov 20, 2026

@spideriq/mail-skills v0.17.3 — Visual card kit and launch film

I updated @spideriq/mail-skills to v0.17.3 on OPVS with a 4-part chalkboard card kit and refreshed first-person listing copy, completing the v0.17.1 62-second launch film so you can see how your agent runs real mailboxes on SpiderIQ before installing. What's new in v0.17.3 Visual marketplace card kit. I added the chalkboard switchboard hero and three workflow panels showing clean YAML thread intake (~37x fewer tokens) with document conversion, coordinated reply and campaign rails, and guarded multi-mailbox operation. 62-second chalkboard walkthrough on the card. Carries forward v0.17.1's narrated launch film on media.demo_video_url, walking through how SpiderMail replaces single-message 1990s webmail across hundreds of accounts. Refreshed 88-method listing guide. I updated marketing.long_description and five verb-led feature cards to reflect all 88 client methods across inbox threads, Jinja2 templates, cold sender warmup, BYO-domain broadcasts, and human-activated Class-B sequences. Every real send stays gated behind get_job_status delivery polling, inbound prompt-injection quarantine, outbound credential scanning, and automatic pause above 0.3% complaints. Trigger it with "read my inbox", "reply to this thread", or "send a broadcast". Live on the marketplace as @spideriq/mail-skills.

OPVSRead →
mail-skills-v0.17.2Nov 20, 2026

@spideriq/mail-skills v0.17.2 — Visual card kit and launch film

I updated @spideriq/mail-skills to v0.17.2 on OPVS with a 4-part chalkboard card kit and refreshed first-person listing copy, completing the v0.17.1 62-second launch film so you can see how your agent runs real mailboxes on SpiderIQ before installing. What's new in v0.17.2 Visual marketplace card kit. I added the chalkboard switchboard hero and three workflow panels showing clean YAML thread intake (~37x fewer tokens) with document conversion, coordinated reply and campaign rails, and guarded multi-mailbox operation. 62-second chalkboard walkthrough on the card. Carries forward v0.17.1's narrated launch film on media.demo_video_url, walking through how SpiderMail replaces single-message 1990s webmail across hundreds of accounts. Refreshed 88-method listing guide. I updated marketing.long_description and five verb-led feature cards to reflect all 88 client methods across inbox threads, Jinja2 templates, cold sender warmup, BYO-domain broadcasts, and human-activated Class-B sequences. Every real send stays gated behind get_job_status delivery polling, inbound prompt-injection quarantine, outbound credential scanning, and automatic pause above 0.3% complaints. Trigger it with "read my inbox", "reply to this thread", or "send a broadcast". Live on the marketplace as @spideriq/mail-skills.

OPVSRead →
publish-skills-v0.45.2Nov 20, 2026

@spideriq/publish-skills v0.45.2 — Sites, funnels, and booking

I shipped @spideriq/publish-skills v0.45.2 on OPVS with a 59-second chalkboard explainer film and a 4-part visual card kit so you can inspect the full SpiderPublish architecture right on the marketplace listing. Instead of gluing five human page, form, funnel, and calendar apps together or paying a web agency retainer, your agent builds, themes, and deploys complete client websites from one chat. What's new in v0.45.2 59-second explainer film on the listing. Watch the narrated chalkboard walkthrough directly via media.demo_video_url (added in v0.45.1 and carried forward in v0.45.2), showing how one agent replaces five disconnected web tools. 4-part visual marketplace card kit. I added the full-stack workspace hero and three chalkboard panels covering DIY-versus-agency replacement, the built-in template and conversion marketplace, and two-phase edge verification. Refreshed first-person listing guide. I updated marketing.long_description and five verb-led feature cards covering 172 SpiderPublish HTTP methods, programmatic SEO directories, /lp/ prospect pages, branching funnels, and /f/<flow_id> forms and booking. Two-phase edge safety and live visual checks. Every broad mutation stays gated behind get_auth_status, dry_run + confirm_token, and content_visual_check over dom.shadow_hosts. Trigger it with "build a landing page", "create an SEO directory", or "set up a booking funnel". Live on the marketplace as @spideriq/publish-skills.

OPVSRead →
agent-ops-skills-v0.11.1Nov 20, 2026

@spideriq/agent-ops-skills v0.11.1 — Verified agent loop + film

I updated @spideriq/agent-ops-skills to v0.11.1 with a 70-second chalkboard explainer film and a 4-part visual card kit on OPVS, pairing v0.11.0's dual Claude Code and Codex command renderer with SpiderIQ's measured software delivery lifecycle. What's new in v0.11.1 70-second explainer film and visual card kit. Watch the narrated chalkboard walkthrough and inspect 3 workflow diagrams directly on the marketplace card, restoring demo_video_url from v0.10.1 alongside the new hero and panels (v0.11.1). Dual render targets for Claude Code and Codex. Set project.harness: [claude-code, codex] in ops/profile.yaml so render-commands.sh emits native Codex skills, checks AGENTS.md byte caps, pages rulebook reads, and runs isolated sessions via codex-session.sh (v0.11.0). Zero-guess repository discovery and proof. Run discover-profile.sh to keep unreadable settings marked MISSING, prove-profile.sh to test each field against a probe and a negative control, and render-commands.sh --require-proven to render 14 byte-checked commands. Script-enforced ledgers across three roles. Separate your Planner (checklists derived from 402 real cards across 8 boards), isolated Worker (/start, /test, /ship), and Orchestrator deploy drain behind PASS, FAIL, and CANNOT exit codes where an unmeasured check never counts as passing. Your workspace can now adopt a proven engineering loop across both Claude Code and Codex without hiring a systems architect. Live on the marketplace as @spideriq/agent-ops-skills. Trigger it with "adopt agent-ops into this repo".

OPVSRead →
spiderpr-skills-v0.1.2Nov 18, 2026

@spideriq/spiderpr-skills v0.1.2 — Put your news on the wire

I updated @spideriq/spiderpr-skills to v0.1.2 with a 53-second chalkboard explainer film and a 4-part visual card kit on OPVS, giving your agents a clear guide for distributing approved company press releases over the external newswire through SpiderIQ's SpiderPR pipeline. What's new in v0.1.2 53-second explainer film and visual card kit. You can now watch the narrated chalkboard walkthrough and inspect 3 workflow diagrams right on the marketplace listing before installing (v0.1.1 and v0.1.2). One-call async wire submission. Your agent calls submitPressRelease (POST /jobs/spiderPR/submit) with your approved title (up to 300 chars) and body (up to 50,000 chars), plus optional summary, category, 25 tags, and press contact. Embargo scheduling and safe test runs. Set payload.scheduled_release_at as an ISO 8601 timestamp for timed launches, or pass payload.test: true to verify the submission flow on the test queue before spending credits. Anti-double-spend polling and live wire receipts. A confirm-before-you-wire gate prevents blind retries on paid jobs while getJobStatus and getJobResults return your live published_url and wire_report_url. Your press desk now pairs external wire pickup cleanly with your own newsroom (@spideriq/publish-skills) and outreach (@spideriq/mail-skills). Live on the marketplace as @spideriq/spiderpr-skills. Trigger it with "send this press release".

OPVSRead →
spiderflows-v0.13.2Nov 17, 2026

@spideriq/spiderflows v0.13.2 — Live lead discovery

I shipped @spideriq/spiderflows v0.13.2 on OPVS so you can watch the 57-second chalkboard walkthrough and inspect the full visual architecture right on the marketplace card before your agent runs a single SpiderIQ pipeline. Instead of renting stale database exports or wiring five separate scrapers by hand, your agent dispatches live discovery, website and social crawling, registry lookups, LinkedIn extraction, and SMTP email verification straight into IDAP. What's new in v0.13.2 Visual marketplace card kit. I added the chalkboard dispatch hero and three architecture panels showing 470-slot parallel worker fan-out across 11 VPS, pre-flight cost and corpus dedup gates, DIY-versus-agency replacement, and verified IDAP contact cards. Embedded 57-second chalkboard walkthrough. Carries forward v0.13.1's narrated launch film directly on media.demo_video_url, walking through how an agent replaces decayed Apollo spreadsheets with live discovery and deliverability checks. Refreshed first-person listing guide. I updated marketing.long_description and five verb-led feature cards covering single-state US ZIP campaigns, Sortlist's 109 agency services across 21 countries, internal corpus re-enrichment, and Company Intel briefs for up to 50 companies. Every campaign still gates spend first with a free read-only /estimate check and quotes run duration from my account's jobs_per_hour quota. Trigger it with "find local businesses in Austin", "find marketing agencies in Germany", or "research this company". Live on the marketplace as @spideriq/spiderflows.

OPVSRead →
media-skills-v0.2.2Nov 17, 2026

@spideriq/media-skills v0.2.2 — Reuse every hosted asset

I shipped @spideriq/media-skills v0.2.2 so your OpenClaw agents can find, inspect, and reuse every image, video, and PDF your SpiderIQ workflows already hosted in SpiderMedia across SeaweedFS, Cloudflare R2, and PeerTube. Instead of copying bucket links by hand or paying a model to generate duplicate visuals, you give every workflow one read-only catalog lookup and watch the full walkthrough right on the marketplace listing. What's new in v0.2.2 Chalkboard explainer film on the listing. You can now watch a 55-second narrated walkthrough directly on the @spideriq/media-skills page showing how listAssets, searchAssets, and getAsset connect your three storage tiers into everyday agent workflows. Visual marketplace card kit. I added the chalkboard manifold hero and three system panels covering three-station lookup (listAssets, searchAssets, getAsset), server-side SPIDERIQ_PAT workspace isolation with 40% to 76% YAML or Markdown token savings, and find-before-create asset reuse. Refreshed first-person listing guide. The marketplace card now spells out the exact AND versus ANY tag filter rules, catalog UUID (id versus peertube_uuid) lookups, and how @spideriq/media-skills pairs with producer skills like @spideriq/publish-skills and @spideriq/spiderflows. Trigger it in any workflow with "what media do I have" or "search the media library". Live on the marketplace as @spideriq/media-skills.

OPVSRead →
executor-v1.0.1Nov 13, 2026

@opvs-ai/executor v1.0.1 — See the work behind the outcome

Your agent needs the outcome, not every API response. I added Executor's launch film to its marketplace listing in v1.0.1. It shows how supported jobs run on the server and save findings to your OPVS CRM board. The skill's behavior is unchanged. What's new in v1.0.1 Watch the workflow on the listing. The launch film follows a job from the agent's request to saved board findings. It explains why bulk results can go straight to the board while the agent keeps room for your next instruction. See what the agent still handles. In the existing workflow, your agent submits a supported task or board job, receives a job ID, and checks progress. The server fetches data, runs the skill, and writes results back. This release adds the film, with no change to those steps. Choose when findings enter the conversation. The film shows the same boundary as Executor's existing guidance: use it for bulk work that can run outside the conversation. Bring the findings into context when the agent needs to read them and reason about the result. The film is available on the marketplace card. Start with a supported skill and an explicit source column when running a board job. Live on the marketplace as @opvs-ai/executor. Trigger it with "enrich board" or "run enrichment".

OPVSRead →
firecrawl-extract-branding-v1.1.0Nov 13, 2026

@opvs-ai/firecrawl-extract-branding v1.1.0 — A checked visual brief

Your new client sent a website. The brief is blank. I want the next page to look like their business before design starts. Firecrawl extract branding v1.1.0 gives the agent a clearer route from that page to a checked visual profile. What's new in v1.1.0 Find the skill under its Firecrawl name. Install @opvs-ai/firecrawl-extract-branding, previously listed as @opvs-ai/extract-website-branding. The former name remains a search alias. The extractBranding method and its parameters stay the same, so the extraction task is familiar. Build the brief from what the page returned. The guidance covers colours, font names, logo, spacing and button styles, plus the returned brand name and confidence. It includes saved examples from five public sites, with recipes for onboarding, matching a design and comparing brands. Keep warnings beside the visual references. The bundled checker tests returned colours for contrast and labels a profile with overall confidence below 0.8 as a draft. A colour that fails contrast stays decorative; it should not become the next page's text colour. One call reads one page and uses one Firecrawl credit. The profile describes that page; font names are references, not licensed font files. The team still owns the design decisions. Live on the marketplace as @opvs-ai/firecrawl-extract-branding. Trigger it with "brand colors" and the client's website.

OPVSRead →
agents-v1.0.1Nov 13, 2026

@opvs-ai/agents v1.0.1: HR for your agents

Your next agent needs onboarding before it can work. I have added the Agents launch film to its marketplace listing. It follows a customer enquiries role through the setup: choosing skills, writing instructions and deciding who the employee reports to. What's new in v1.0.1 Watch the onboarding story. The listing now carries the launch film. It shows how your existing agent can use Agents as HR for an AI team, from choosing an employee to setting up its role. The story illustrates the workflow; it does not report a measured hiring outcome. See what each hire needs. The film walks through skills, tool guidance, persona and instructions, then managers, working rules and reports. These capabilities already belong to Agents. Hiring needs your approval, and changes must stay within the caller's permissions. Some vendor-owned files are locked. Keep the skill behavior you use. Version 1.0.1 adds the film link and release note. It changes no skill methods. The existing skill brings together hiring, persona files, reporting lines, AI Office and website chat. The employee still needs the tools and access its job calls for. Live on the marketplace as @opvs-ai/agents. Trigger it with "Help me hire and set up an AI employee for customer enquiries."

OPVSRead →
reminders-v0.1.0Nov 13, 2026

@opvs-ai/reminders v0.1.0: Wake with a purpose

An agent that wakes every few minutes just to check a pending approval spends tokens before the next job can begin. I shipped Reminders so you can give that agent a reason to return: a follow-up tomorrow, a weekly review, or a wake when an approval card is done. What's new in v0.1.0 Breaking: use the Reminders package name. Reminders replaces @opvs-ai/calendar. Existing integrations using that package's calendar_* methods need the new reminders_* names. The four native OpenClaw calendar tools keep their names and use the same calendar. Schedule the return for real work. Give the agent a time, an interval, a cron schedule, or a supported condition such as a card reaching done. Repeats have a chosen end; you can pause, resume or cancel an entry as plans change. Keep empty checks off the model bill. A clock checks supported conditions without calling a model. In live mode, it can wake the agent with the task to resume, within your policy. Quiet hours and limits still apply. Shadow mode records what would happen without waking anyone. Check before relying on a wake. Preview an entry without saving it, test a supported condition, and read the wake history with outcomes and costs. Find @opvs-ai/reminders on the marketplace. Trigger it with "remind me tomorrow at 9" and include your time zone.

OPVSRead →
code-index-v0.7.1Nov 13, 2026

@opvs-ai/code-index v0.7.1: Give your coding agent eyes

Your agent can code. Must it read every line? I want it to find where a task belongs before it writes a fix or builds a feature. Code index gives it a map of the indexed repository: files, functions and the connections it can read. What's new in v0.7.1 Watch the launch film on the store card. This patch adds the film to the marketplace listing. It shows the existing workflow through a feature request: find a report builder, inspect the matching function and check where it is used. Check before writing. The live skill lets an agent search for code that already exists, then read the relevant files and symbols. It helps the agent decide what to reuse or change. The coding agent still does the implementation. Read the limits with the map. Indexed callers cover Python and can miss calls. Import graphs cover Python and TypeScript/JavaScript. Check the indexed commit, age and truncation before trusting an answer. Test suggestions help choose what to run first; they do not prove coverage. If the index is unavailable, the agent can continue by reading the repository directly. Live on the marketplace as @opvs-ai/code-index. Trigger it with "code index" and give it the project's AgentBoard project ID.

OPVSRead →
opvs-communication-v1.0.2Nov 13, 2026

@opvs-ai/opvs-communication v1.0.2: One shared language

I want agents in different apps to pass work directly, with the context attached. OPVS Communication gives each connected agent an address and a shared way to ask, assign, hand off and reply. The v1.0.2 listing adds the approved launch film to show that exchange. This patch changes no skill behavior. What's new in v1.0.2 Watch the handoff. The marketplace now includes the launch film. Its example puts Gemini on research, Codex on the build and Claude Desktop on review. That is an illustrative team, not a claim that every app connection was tested. See who receives the work. The film shows addresses for connected agents. A research question, a build request and a review handoff each carry a clear intent. Each agent keeps its own tools. Follow the reply. The story shows the recipient replying in the same thread, so the request and its context stay connected. These mail and message-format tools already belong to the merged skill; this patch adds the film. The merged release also includes team rooms and DMs. Use mail for a direct handoff and a room for a team discussion. It brings together opvs-protocol, opvs-messaging, opvs-parser and agentrooms, keeping their method names. Find it on the marketplace as @opvs-ai/opvs-communication. Trigger it with "Ask the reviewer agent to check the release notes."

OPVSRead →
agentdocs-v1.2.1Nov 13, 2026

@opvs-ai/agentdocs v1.2.1 — Watch the launch film

I’ve added the AgentDocs launch film to its marketplace listing. It follows one client launch: a date in a chat, tasks on a board and a plan in another file. The film shows how your agent can gather that context into project pages you can read and update together. Watch the AgentDocs launch film. What's new in v1.2.1 Watch the workflow before you install. The listing now includes the launch film. Follow the example from scattered project context to Markdown pages, then see how an agent can find the plan and write an update. You can watch it on the skill page. See where access and history fit. The film explains sharing with authorized people and agents, reviewing changes and restoring an earlier version while retaining history. These are existing AgentDocs capabilities; the example gives them a concrete place in a project handoff. Keep the skill behavior you already use. Version 1.2.1 adds the film link and release note. It changes no methods, permissions or page workflows. Projects, search, revisions and rollback continue to work as in 1.2.0. Live on the marketplace as @opvs-ai/agentdocs. Trigger it with "Find the project plan in AgentDocs and read it back." View AgentDocs.

OPVSRead →
v1.15.1Nov 12, 2026

@opvs-ai/agentmemory v1.15.1: Shared memory across agents

Your AI needs knowledge. Must every prompt carry it all? When you plan a new campaign, pasting the whole chat history bloats context. Cutting it leaves your next agent missing key facts, so someone rebuilds notes by hand. I shipped @opvs-ai/agentmemory v1.15.1 with the 101-second launch film on opvs.ai/skills/agentmemory, right on top of v1.15.0's full route coverage. What's new in v1.15.1 Launch film on the store page. Version 1.15.1 embeds my 101-second film on the listing. You can watch how your brand's agents save shared knowledge once and pull back only what the next task needs. Every agent route is a method. Version 1.15.0 added 25 methods for 47 in total. Your agents can call listMemories, getMemory, source sync (listSources to syncSource), session and run logs, exportMemories, listDeskMemories, syncKnowledge, and profession reads, with every route field exposed. Five memory kinds and clean edits. Agents save a learning, decision, pattern, preference, or fact. Before writing, an agent checks existing entries and calls updateMemory or deleteMemory instead of adding a conflicting copy. Task recall and shadow review. Another agent in your brand runs searchMemories for exact phrases or recallRelevant for ranked context, then checks facts that may have changed. Code lessons (entriesNear, anchorsLookup) and shadow verdicts (listMemoriesForReview, submitMemoryVerdict) stay built in. Live on the marketplace as @opvs-ai/agentmemory. Trigger it with "remember this decision" or "recall what my agents know".

AgentMemoryRead →
blotato-v2.1.1Nov 12, 2026

@opvs-ai/blotato v2.1.1: Give your agent the publishing job

Your business needs you, and your posting needs an agent. Even when posts are ready, someone still has to pick accounts, set times, and come back to check what happened. I can do that recurring job myself, or pay someone else to carry it. Now my agent handles publishing across X, LinkedIn, Facebook, Instagram, TikTok, Pinterest, Threads, Bluesky, YouTube, and Slack. I supply the finished posts, connected accounts, and timing, then review the outcomes. Watch the 50-second launch film at https://videos.spideriq.ai/videos/embed/cf85b398-67ff-4ac7-aa54-07dc935f1ab4?title=0&warningTitle=0&peertubeLink=0&p2p=0. What's new in v2.1.1 New @opvs-ai/blotato identity and launch film. Previously published as @opvs-ai/vibe-publish, the skill now ships under @opvs-ai/blotato. It keeps vibe-publish as a search alias, routes per-brand Blotato keys in Settings, and embeds the launch film on the store page. Destination lookup for Pages, channels, and boards. My agent calls listSubaccounts, listPinterestBoards, and listAccounts with a platform filter. It finds Facebook Pages, LinkedIn company Pages, Slack channels, YouTube playlists, and Pinterest boards without asking me for raw IDs. Outcome verification with getPostStatus. An accepted send is not proof that a post is live. getPostStatus polls Blotato and reports whether each destination is published, scheduled, failed, or still processing. Pre-send payload gates and complete target schemas. Two bundled scripts (verify-post-payload.mjs and verify-published.mjs) check character limits, media counts, and platform rules before sending. The schema also fixes Instagram target.trial and declares nine TikTok, YouTube, and Slack target fields. Live on the marketplace as @opvs-ai/blotato at https://opvs.ai/skills/blotato. Trigger it with "publish this post", "schedule this thread", or "check if my post went live".

OPVSRead →
v1.17.1Nov 11, 2026

OPVS v1.17.1 — the client kit routes an idea to a board

The OPVS client kit and the @opvs-ai/opvs skill, now 1.3.1, give a new client's coding session a route for an idea. In our own test, a fresh session with the old kit found no such route, so it built a web page and opened no board. What's new in v1.17.1 An idea has its own row at the front door. The first row of the skill's "What are you trying to do?" table is now "I have an idea, build me X". Its recipe confirms the idea with one question, plans it, makes the board, writes cards with dependencies and opens the first card's pull request in your own repository. Your editor lists the OPVS skills. The kit now writes a short header, a name and a description, on every skill file it copies, so Claude Code lists them. Before, it copied them without one, and our test session saw none of the 9 OPVS skills. The project block names only what the kit ships. The preflight check runs with --online, and staff commands such as /start and /ship are marked "OPVS staff only", because a client install does not include them. Onboarding says what you do first. Your brand admin switches agent tokens on before handing over the kit, you approve the project's MCP server when you first open it, and a machine without unzip extracts it with python3 -m zipfile -e. Every step a new client needs is now written inside the kit itself. Read the skill at https://opvs.ai/skills/opvs-ai/opvs and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.17.0Nov 10, 2026

OPVS v1.17.0 — builds that fail only for the code

Factory builds now fail for the code's reasons, not the machine's. A machine fault no longer costs a card a try, every failure names its cause, builds get room to finish, and you can read and message a build while it runs. What's new in v1.17.0 A fault on our side costs the card nothing. When the machine fails a build (an expired sign-in, a refused key, a crashed coding tool, a box too busy to test), the card keeps its try and its budget, and another machine picks it up. Every failure names its cause. GET /api/v1/factory/ratings/causes lists each model's failures by cause and keeps machine faults in their own env_faults count, so a machine fault never lowers a model's pass rate. Builds get room to finish. Every builder now runs up to 230 minutes, and a builder that counts steps gets 600. A busy build is never stopped as stalled, and a build that reaches its limit is paused with its work saved. Read a running build and talk to it. opvs builds read shows a running build's current command, how long it has run, its last steps and its messages. opvs builds tell sends the build a message, or saves it on the card if the build has ended. A desk settings change now reports whether the machine applied it, and answers "not applied" with the reason when it did not. On our own factory, failed builds fell from 38 to 22 per 100 in the week after these changes. Read how a build fails and recovers at https://opvs.ai/docs/factory/fix-a-build and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.16.0Nov 7, 2026

OPVS v1.16.0 — browse every skill on opvs.ai, sign in to download

The skills store on opvs.ai/skills is open to everyone. Each public skill has a page that says what it does, with its videos, articles and the AI tools it works with. Downloading or installing a skill needs a free OPVS account. What's new in v1.16.0 Breaking: downloading a skill package now needs a sign-in. The registry's download and skill-guide routes answer 401 to a request without a token, and package lists no longer carry a direct file address. To download from a script, send your OPVS personal access token as Authorization: Bearer. Anyone can read a skill before signing up. Every public skill has its own page on opvs.ai/skills, with its description, videos, articles, the AI employees that use it and more from its vendor. Browse by category, hashtag or vendor, or search from the header on any page. From a skill page to your own tools in one click. Download a skill for Claude Desktop, Claude Code, Codex or Antigravity. Signed out, a free sign-in brings you straight back to the download. Install to workspace opens the dashboard's install wizard, which names the workspace it installs into. One list behind the site and the dashboard. opvs.ai/skills and the dashboard marketplace read the same registry, and OPVS sets where each package appears. A team member who is not a brand admin sees Ask your admin to install, and every skill an agent lists links to its marketplace page. Read how the store works at https://opvs.ai/help/give-your-agents-skills/browse-the-skills-store and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.15.0Nov 7, 2026

OPVS v1.15.0 — see which desks can be given work

The fleet read now says what role each desk takes work under, and it flags a desk that is taking work with no role. A board that hands cards only to desks with a role would give such a desk nothing, and until now nothing said so. What's new in v1.15.0 Every desk shows the role it claims work under. GET /api/v1/environments returns worker_role for each desk, such as builder. The desk reports it on every health check, so the value is what the desk is running now, not what it was set up with. A desk that is taking work with no role is flagged. no_role is true only when the desk is running its claim loop, is not dormant and reports no role. A desk with its loop off is never flagged, so a parked desk does not show as a problem. You can tell a running desk from an idle one. worker_loop_running says whether the desk is claiming cards. A null means OPVS has not measured it yet, which is not the same as false. The same answer in the IDE. The MCP tool fleet_list shows all three fields on every slot from @opvs-ai/mcp 2.13.4. A builder hired from the catalog gets the role builder set for it. On a desk you run yourself, set OPVS_WORKER_ROLE in the sidecar's settings and restart it. Read the reference at https://opvs.ai/docs/opvs/employees/workstations and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.14.0Nov 1, 2026

OPVS v1.14.0 — ask the code index who calls a function

The code index now answers who calls a function. Name a file or a symbol, and OPVS reads its stored map of your repository for the callers, the test files nearby and the commit that answer describes, with no checkout and nothing run. What's new in v1.14.0 One call shows who depends on a function. POST /api/v1/code/{project_id}/callers takes up to 50 files or symbols. Each caller comes back with the symbol it calls and a label saying whether the match is exact or heuristic. Python calls are indexed today. The same answer in the terminal, the IDE and the agent. Use opvs code callers in @opvs-ai/cli 0.27.9 or later, the MCP tool code_callers in @opvs-ai/mcp 2.12.0 or later, or the callers method in the @opvs-ai/code-index skill. A draft card gets checked before work starts. opvs code check-card <brief.md> reads the paths in a brief and prints one block to paste into the card, with a row per path. A path that does not exist shows as NO while it is still cheap to fix. Unknown never reads as zero. A file type with no call data, a folder or an unavailable index comes back marked unknown with its reason, so an empty list means no caller was found, not that nobody looked. Three limits apply to every answer. The index is as old as its last commit, a heuristic caller may be wrong, and a test hint is not coverage. The route needs a token with the code:read scope and a GitHub repository attached to the project. Read the reference at https://opvs.ai/docs/opvs/api/code-index and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.13.0Oct 30, 2026

OPVS v1.13.0 — dev agents join the Agents page

The Agents page in the OPVS dashboard now lists coding agents beside the agents you hired or created, and every agent has Skills, Work and Settings tabs. A hired agent's vendor files are now protected, with a copy you can change. What's new in v1.13.0 Breaking: the vendor's four files on a hired agent are read only. SOUL, AGENTS, IDENTITY and TOOLS belong to the vendor; USER, MEMORY and HEARTBEAT stay yours. To change the vendor's files, click Make my own copy: it creates a separate agent of your own with all seven files, and the hire stays as it was. Dev agents live in the same list. Coding agents on your machines appear in a Dev agents group with their Terminal, Files, Skills, Work and Settings tabs. A hired dev agent also gets its own mailbox and Memory, and Machines now shows a short row that opens it here. Skills, Work and Settings for every agent. Skills separates the vendor's packages from the ones you added. Work shows the cards assigned to the agent and its recent runs. A hired agent's Settings holds who it reports to, its trust policy, its channels and its subscription. Create your own agent asks for a name, a role and a model. The agent is saved as a draft. Its Settings has six blocks: Identity, Instructions, Skills, Model, Channels and Danger, and only a workspace admin can deploy it. When a vendor ships a new version of a file, Review update shows it and Apply update puts it live without touching your files. Read how the Agents page works in the docs at https://opvs.ai/docs/opvs/dashboard/agents

OPVSRead →
v1.12.0Oct 27, 2026

OPVS v1.12.0 — ask which tests a change can break

The code index now answers which tests a change can break. Send the files you changed, and OPVS walks its stored map of your repository back to the test files that reach them, with no checkout and nothing run. What's new in v1.12.0 One call lists the tests behind a file. POST /api/v1/code/{project_id}/tests-for takes up to 200 repo-relative paths and returns, for each one, the test files that reach it, nearest first, with how many steps away each test is and how it gets there. Tests that read a file count too. A test can reach a file by importing it, or by naming it as a path it opens at run time, such as a fixture. Both kinds of link are followed, up to six steps deep. An unmeasured answer never reads as "no tests". A path the index does not know comes back marked unknown. When there is no ready index, or the index is partial, the answer lists the reason first, so a client knows to run everything instead. The same answer from the terminal. opvs code tests-for --remote --project <id> <paths> in @opvs-ai/cli 0.27.6 or later prints each file with its tests, or the raw answer with --json. The route needs a token with the code:read scope and a GitHub repository attached to the project. Read the reference at https://opvs.ai/docs/opvs/api/code-index and every release note at https://opvs.ai/changelog.

OPVSRead →
v1.11.1Oct 18, 2026

OPVS v1.11.1 — OpenCode and Pi builds use each model's own limits

OpenCode and Pi are two of the coding engines that build cards on an OPVS desk. Before version 1.11.1 both read one context window and one output limit for the whole desk, measured for MiniMax-M3 only. Each build now uses the limits of the model it runs. What's new in v1.11.1 Breaking: a build with unknown limits stops and says why. When a build's model publishes no limits and is not the model the desk-wide values describe, the desk no longer guesses. The build stops before the engine starts and records the failure cause harness_limits_unknown. Pin such a card to a model that publishes its limits. Each build gets its own model's window and output limit. The credential a desk receives for a build now carries that model's context window and output limit. OpenCode and Pi compact the conversation and size each reply from those numbers, so a long build on a 262,144 token model no longer plans for a 1,048,576 token window and fails half-way. A reply always fits inside the window. Some models count the requested output inside the context window. The output limit is now capped at half the window less a 16,384 token margin, and compaction keeps that much room free, so a late turn is not refused for being too long. Proven on a live build. On 18 October an OpenCode build on Kimi K2.7 Code ran with a 262,144 token window, a 114,688 token output limit and 131,072 tokens held back for compaction. Those are Kimi's own limits, not MiniMax-M3's. Desks get these changes with sidecar 0.2.130 or later. Read every release note at https://opvs.ai/changelog.

OPVSRead →
v1.11.0Oct 18, 2026

OPVS v1.11.0 — your own mail address, and Mail to read it

Every person in an OPVS workspace now has their own address at opvs.run, the same kind of address agents use. Agents and people write to it, and you read and answer in Mail, a new item in the dashboard sidebar. What's new in v1.11.0 An address made for you. OPVS makes it when you join a workspace, from your first and last name, such as jordan-lee@opvs.run. A taken name gets a number on the end. You can change it under Settings, Profile, and the old address keeps forwarding for 30 days. Mail in the dashboard. Read your conversations, reply, and write new messages to agents and people. Search covers the whole mailbox, and you can mark a message unread or give it your own labels, which only you see. One mailbox for every workspace you belong to. Agents and people in each of your workspaces can write to you, and their messages arrive in one place. Several sign-ins of the same person can share one address. Your mail from code. The same mailbox answers at /api/v1/me/mail and in the CLI as opvs my-mail. A retry that carries the same Idempotency-Key is delivered once, so pressing Send twice sends one message. Your address does not use an agent seat. Read how your address works in the docs at https://opvs.ai/docs/opvs/tools/your-opvs-address

OPVSRead →
v1.10.0Oct 17, 2026

OPVS v1.10.0 — a roomier Agents page, with favourites

The Agents page in the OPVS dashboard has a wider, easier to read list of your agents and a calmer middle section. The Org chart, which used to sit in a bar above the page, now has its own tab on Home. What's new in v1.10.0 Agents are easier to find and tell apart. The list is 320 px wide, up from 232 px, and every face is 44 px and in colour instead of grey. Names and titles are larger, and a small badge says whether an agent is hired or one of your own. Your most used agents stay at the top. Click the star on any agent and it moves into a Favourites group above the rest of the list. Stars are kept in your browser, for you and for this workspace. The search field opens its own filters. Click it to narrow the list by kind, such as hired or your own, or by state, working or idle. A filter that is on shows as a small tag inside the field, with an × to clear it. Chat, Mail, Memory and Workspace read like the tools you know. The tabs are the first line of the page. Chat is one centred conversation with starter prompts, Mail lists conversations on the left and opens one on the right, and Memory and Workspace are plain tables. The Org chart tab works for every member, and its old link still opens it. Read how the Agents page works in the docs at https://opvs.ai/docs/opvs/dashboard/agents

OPVSRead →
v1.9.0Oct 17, 2026

OPVS v1.9.0 — move or copy a card to another board

A card filed on the wrong board, or a board being split in two, no longer means retyping the card by hand. A card can now move to another board with its whole history, or be copied there as a new card. What's new in v1.9.0 Move a card and keep everything on it. opvs tasks transfer OPS-142 --board WEB moves the card itself: its comments, history, links and pull request go with it, it takes a number on the new board, and the old number still opens it. Copy a card when the same work is wanted twice. opvs tasks copy makes a new card with its own number on the target board, linked to the original, which stays where it is. The copy carries the text, brief, labels and fields, never the comments or the assignee. Agents can do both. The agentboard skill 2.2.8 adds the transferTask and copyTask methods, so an agent, or an editor connected through the OPVS MCP server, can move or copy a card in one call. Nothing lands in Ready by accident. Neither a move nor a copy puts a card where a builder can claim it unless the caller names that column, and both refuse a card that a build is working on. The same two calls work over REST, in the CLI from version 0.27.5, and for agents. Read what each one carries and every refusal in the docs at https://opvs.ai/docs/opvs/api/move-and-copy-cards

OPVSRead →
v1.8.0Oct 16, 2026

OPVS v1.8.0 — a builder's failed fix round keeps its work

A builder that fixes its own pull request after review no longer loses that work when its tests go red. Risky cards can now ask for a hard profile with more time, and a long build stays on the desk that started it. What's new in v1.8.0 A red fix round stays on the pull request. When a card goes back to its builder and the new commits fail the test gate, they are pushed to the same pull request labelled hold:red, not left on the desk. The next attempt and the reviewer start from them, and the label comes off after a green push. A hard profile for risky or large cards. Every coder now has a .hard profile, such as codex.hard or claude.hard: the highest effort and largest turn or step guard that coder takes, and 180 minutes instead of 120. Name it in the card's agent_context.profile; the coder-profiles API lists it. A long build keeps its desk. A build still running after 150 minutes used to look abandoned, so its card was handed to another desk and built twice. The desk now refreshes a live build's card every 30 minutes, so a hard build runs to its own limit in one place. The build record names the right board and session. Each build in the run record now carries the board of the card it built, not the desk's home board. Codex and Claude builds record the same session id as their live events, so a build joins its live log. Desks get these changes with sidecar 0.2.153 or later. Coder profiles, the hard tier included, are documented at https://opvs.ai/docs/opvs/api/coder-profiles.

OPVSRead →
v1.7.0Oct 15, 2026

OPVS v1.7.0 — a lesson can name the test that checks it

A lesson your team writes in the repository can now name the test that checks it. OPVS then says when that test is gone, and an agent about to change some code reads the nearby rules that no test will catch first. What's new in v1.7.0 A rule can point at its test. Add a line Checked by: under a lesson in a knowledge file such as LEARNINGS.md, naming a test file or one test function. A line Derived from: records the incident or card the rule came from. You hear when a rule loses its test. When every test a lesson names is deleted, the next index marks the lesson unguarded. One search lists all of them, so each rule gets a new test or a deliberate decision to drop it. Renaming a test does not raise a false alarm. A renamed or moved test keeps guarding its rule, because it still runs. The lesson shows both the path you wrote and where the test lives now. The riskiest rules come first. Among lessons the same distance from the code being changed, a rule whose test sits near that code now ranks after one with no such test. Opening a test file also lists the rules it protects. Agents reach this through the agentmemory skill, version 1.13.1. Read the line format and the API fields in the docs at https://opvs.ai/docs/opvs/api/memory

OPVSRead →
v1.6.0Oct 15, 2026

OPVS v1.6.0 — a shorter sidebar, and one place for every setting

The OPVS dashboard has a new sidebar and a new Settings area. Every workspace now opens on the same seven places to work, and every setting has one page with its own address. What's new in v1.6.0 Seven places to work, with the two most used actions on top. The sidebar lists Home, Tasks, Boards, Agents, Docs, Mail and Rooms. New task and Search sit above them, Search also opens with Ctrl K or ⌘ K, and your latest chats sit under Rooms with Sigal first. Settings live in one place, grouped by who they affect. Click Settings and the sidebar becomes the settings menu: My account for what is only yours, Workspace for what applies to everyone, and Building for workspaces that run coding agents on their own machines. Each page has its own address, such as app.opvs.ai/settings/members. Admins choose the optional features. Workspace, then Features, turns Skills, Feed, Machines and CRM on or off for the whole workspace. Turning one off hides it and deletes nothing: what it holds comes back when it is turned on again. Old links still work. Every old settings address opens its new page and keeps its parameters, so bookmarks and links back from billing or GitHub land in the right place. Every active workspace has used the new sidebar since 11 October. Read where each setting lives in the docs at https://opvs.ai/docs/opvs/dashboard/find-your-way-around

OPVSRead →
v1.5.1Oct 15, 2026

OPVS v1.5.1 — switching brands reloads, and the switch sticks

The brand switcher in the dashboard's user menu decides which brand every page shows. Before version 1.5.1 a switch could quietly undo itself, so people kept working in the brand they had just left. Choosing a brand now reloads the dashboard into it, and it stays there. What's new in v1.5.1 Choosing a brand reloads the whole dashboard. The page loads fresh in the brand you picked, the checkmark moves to it, and every request the dashboard sends from then on carries that brand. You land on a start page, because a board or task from the old brand means nothing in the new one. A switch can no longer undo itself. A page opened from a link to a board, task, agent, hire page or machine follows the brand that link belongs to. It used to pull you back there right after you switched away. Now it follows its link once, when it opens, and after that the brand is yours to choose. Picking the brand you are already in does nothing. No reload happens, so a stray click in the menu costs you nothing. Other open tabs follow without reloading. A dashboard tab you left open moves to the new brand in place, so an unsaved draft in it survives. Linked pages in those tabs no longer switch the brand back either. You see the brand you chose, on every page, until you choose another one. Read every release note at https://opvs.ai/changelog.

OPVSRead →
v1.5.0Oct 11, 2026

OPVS v1.5.0 — see your code, and keep lessons on it

The code index now draws a project's code as a graph and ties each lesson your team writes to the code it is about. An agent about to change a file gets the lessons for that file and its neighbours, not the whole rulebook. What's new in v1.5.0 You can see your code as a graph. A project's Code index has a new Graph view. It opens on folders and the imports between them. Pick a file and it is drawn in columns, with the files that import it on one side and the files it imports on the other. Lessons follow the code they are about. A lesson can be linked to a file, a function or a folder. When that code is renamed or moved, the lesson moves with it. When it is deleted, the lesson is marked stale and shows up in its author's list of stale lessons. Agents get the lessons near the code they will change. Given the files a card will touch, OPVS walks two steps out through imports and folders and returns the lessons linked along the way, pinned ones first, each with how it was reached. Busy shared files are named, not walked through. Fewer copies of the same lesson. Before writing a lesson, an agent can ask whether a similar one already exists. A new lesson that closely matches an old one is saved with a pointer to it, so the two are easy to merge. Agents reach all of this through the code-index and agentmemory skills, the opvs CLI and the MCP server. Read the methods and defaults in the docs at https://opvs.ai/docs/opvs/api/code-index and https://opvs.ai/docs/opvs/api/memory

OPVSRead →
v1.4.1Sep 28, 2026

OPVS v1.4.1 — agents keep what they learn, in the file they read

AgentMemory turns an agent's conversations into long-term memories and writes the best of them into the agent's MEMORY.md. Version 1.4.1 fixes both halves: conversations are picked up on a schedule, and memories reach the file each agent actually reads. What's new in v1.4.1 Conversations become memories without a chat closing. Every 10 minutes the service picks up conversations that are still waiting. A failed model call no longer marks a conversation as done: it is retried after 10 minutes, 30 minutes, 2 hours and 8 hours. Memories land in the workspace the agent reads. The hourly export now finds each agent's real workspace through the gateway's own registry. It writes only its own marked section, and never writes back a file it could not read first, so an agent's own notes stay intact. Correct a memory's summary in place. updateMemory in the @opvs-ai/agentmemory skill 1.3.0 now takes description, the one-line summary recall shows beside a memory's title. Before, the field was accepted and silently ignored. Memory sources stay inside your brand. A memory source can only name an agent your brand owns. Every other agent gets the same 403, whether or not it exists. An agent now carries what it learned from one session into the next. Every memory method, its parameters and its errors are in the docs at https://opvs.ai/docs/opvs/api/memory

OPVSRead →
v1.0.2Sep 28, 2026

@opvs-ai/profession v1.0.2 — names and dates no longer block lessons

The profession skill lets an agent search, pull and contribute to its profession's shared guild hall. Version 1.0.2 brings the contribute guidance in line with the server: fewer lessons are refused outright, and review decides the rest. What's new in v1.0.2 Contribute refuses only details that identify on their own. A credential, an email address, a phone number, an id or an account number gets a 422 and nothing is stored. The error names the class of detail, never the text, so the agent can remove it and send again. A name, an amount, a date or a quote no longer blocks a lesson. The lesson is accepted and stored as pending. The response's flagged list names each detail found, such as body:named_entity, so the agent knows what review will have to remove. Review anonymizes a lesson or seals it for good. Before a pending lesson is shared with the whole profession, review runs the full curation membrane. The lesson is rewritten without the specifics if it can be. If it cannot, it is sealed and never shared. Nothing is shared as sent when the model is out. If the model that anonymizes is unavailable, review answers 503 and the lesson simply stays pending until it can run again. An agent now learns at contribute time what is refused and what review will have to remove before anything is shared. The refusal classes, the flagged list and every error are in the docs at https://opvs.ai/docs/opvs/api/professions

ProfessionRead →
v1.4.0Sep 28, 2026

OPVS v1.4.0 — agents can ask if the code already exists

When a project has a GitHub repository attached, OPVS now keeps a structural index of its code and rebuilds it after every merge. Agents can check what is already there before they write something new. What's new in v1.4.0 Agents can ask whether something already exists. One call searches the project's functions, classes and methods by name and by the first line of their doc comments. It answers exists, similar or none, and names the file and lines of every match. The index follows your merges. A merge to the attached branch queues a fresh index within seconds, and a check every five minutes catches any merge that was missed. Every answer names the commit it describes, so an agent can see how current it is. A build desk starts with a map of the repository. When a repository has no CLAUDE.md of its own, a build desk set up for it gets a generated one describing its layout, languages and test command. A repository that has its own keeps it untouched. Your code is read by a parser, not a model. The index is built on your own workspace's build machines from the repository's files. No source is sent to a model provider, and a failed index posts one in-app notification without stopping a build. The index covers repositories hosted on GitHub, and finds symbols in Python, TypeScript, JavaScript and Go. The routes, limits and error codes are in the docs at https://opvs.ai/docs/opvs/api/code-index

OPVSRead →
v1.3.0Sep 26, 2026

OPVS v1.3.0 — a column can say how long its cards may wait

OPVS boards hold the work that people and agents share. This release lets each column set its own staleness threshold for board health, and makes four board API calls behave the way their documentation says. What's new in v1.3.0 Each column can set how long a card may wait. In Board settings, open a column and fill in Stale after (days) under Board health. The health check then judges that column by its own number, so a Deploy column that holds cards for a week stops reading as stale. Column edits that were ignored now apply. Changing a column's position or slug through the API now takes effect and renumbers its neighbours. It used to answer 200 and change nothing. A slug another column already holds returns 409 and names that column. A card can be moved by column name on every route. Updating a card with a column name now works the same way moving it does. When two columns on a board share the name, the call returns 422 listing both ids and leaves the card where it was. Columns follow their board's access rule. A column on a private board you are not a member of answers 404, exactly like one that does not exist. A malformed board reference on the health routes now returns 422 instead of 500. Scripts that drive boards through the API now get the answer the documentation promised. The routes and error codes are in the docs at https://opvs.ai/docs/opvs/api/board-health

OPVSRead →
v2.2.1Sep 26, 2026

@opvs-ai/agentboard v2.2.1 — health checks you can pin and gate on

The agentboard skill gives an agent typed calls against OPVS boards. Versions 2.0.0 to 2.2.1 publish the list of board health checks, let a column set its own staleness threshold, and make one breaking change to toggleMarker. What's new in v2.2.1 Breaking: toggleMarker no longer takes task_id. The parameter was required but never read. The call flips the marker's own completed state and does not touch any card. Remove task_id from your calls; an install still pinned to 1.x keeps the old signature. The health check set is published. healthCheckContract returns every board and project check with its levels, the parameter that switches it on, a version and a digest. An agent can pin the version and treat any check it does not recognise as a failure. A column can carry its own staleness threshold. createColumn and updateColumn take health_column_age_days, so a Deploy or Test Live column that holds cards for days is judged by its own number. The rest of the board keeps the one boardHealth is called with. List rows say which cards have a brief. Each listTasks row carries has_description and has_instructions, so an agent can find the cards worth opening without calling getTask on every one of them. An agent that gates work on board health can now tell a new check from a passing one. The checks, profiles and parameters are in the docs at https://opvs.ai/docs/opvs/api/board-health

AgentBoardRead →
v1.2.1Sep 20, 2026

OPVS v1.2.1 — a board guest can work it, not rewire it

A brand can invite someone from outside it onto a single board. From v1.2.1 that invitation is scoped more tightly: a guest works the cards at the tier they were given, and the board itself stays with the brand that owns it. What's new in v1.2.1 Board settings and membership belong to the owning brand. Adding or removing members, renaming the board, changing its settings or visibility, and deleting it are now owner actions at every guest tier. A guest who tries one gets the same answer as if the board were not there. The bulk export is an owner action too. Reading a whole board out as one CSV, JSON or Markdown file is reserved to the owning brand. The board room is unchanged: channels, messages, reactions and files stay readable at whatever tier a guest was granted. The protocol cockpit reports your own brand. The admin OPVS page, and the activity, stats and escalation figures behind it, now need a signed in caller and show that caller's own brand. Each brand sees its own agents, its own messages and its own totals. Live connections are accepted only from the OPVS dashboard. The board, docs, chat, terminal and take over connections now check that the page opening them is the dashboard itself. Clients that connect without a browser page, such as the CLI and the MCP server, are unaffected. A brand can now put a contractor on one board without handing over the board. The guest tiers and the connection rules are documented at https://opvs.ai/docs/opvs/api/authentication

OPVSRead →
v1.36.0Sep 20, 2026

@opvs-ai/agentboard v1.36.0 — every call names who may make it

The agentboard skill gives an agent typed calls against OPVS boards. From v1.36.0 five of those calls state which brand is allowed to make them, so an agent learns the rule when it reads the method instead of discovering it from a refusal. What's new in v1.36.0 The four board write calls name their owner. updateBoard, deleteBoard, addBoardMember and removeBoardMember now carry the same owner-only note the column and project calls already had. A guest invited onto another brand's board gets a 404 on all four, at every guest tier. A refused member call leaves nothing behind. addBoardMember states that the ownership check runs before any member record is written. An agent whose call is turned down can retry against the right board without first cleaning up a half-created row. updateBoard names what a guest will actually notice. Shared table column toggles and custom fields are both stored on the board itself, so a guest seat cannot save either one. The method says so rather than leaving an agent to read the refusal as a fault. The board export states who may read it. exportBoardMarkdown now says the bulk export belongs to the owning brand, while board room content stays readable at whatever tier a guest was given. The narrower rule no longer reads as a blanket one. An agent planning work on a shared board can now tell which calls will land before it makes them. The board access rules are documented at https://opvs.ai/docs/opvs/api/authentication

AgentBoardRead →
v1.2.0Sep 13, 2026

OPVS v1.2.0 — Many desks on one machine, only if you say so

An OPVS machine used to carry one desk for one team. From v1.2.0 it can carry several, and it can host a desk belonging to another OPVS customer if its owner switches that on. Each desk is isolated from the others by the operating system. What's new in v1.2.0 Every desk runs as its own user. Two desks on one machine used to share one operating system account, so the file permissions between them protected nothing. Each desk now gets its own account and a private home, and cannot read another desk's stored credentials. One machine, several desks. A machine that carried a single desk can now carry several, each with its own settings, its own port and its own service. Adding the second one needs no new install command and no new token. You decide who may share your machine. Hosting another customer's desk is off on every existing machine, and only the machine's owner can turn it on. Switching it off again stops the next placement and leaves a desk already running in place. The setup command stops instead of overwriting. Running the installer again on a machine that already has a working desk used to replace it. It now refuses, names what it found, and leaves the existing desk alone. A team that needed one server per brand can put those brands on one machine. Machines, desks and the sharing setting are documented at https://opvs.ai/docs/opvs/employees/workstations.

OPVSRead →
v1.1.1Sep 7, 2026

OPVS v1.1.1 — Install works, and the catalog counts are real

A review of the AI Employee catalog turned up six defects, every one of them in a surface a customer touches directly. The marketplace Install button, the hire counts on every product page, paging through admin lists, and two API answers that claimed more than they had done. Version 1.1.1 repairs all six. What's new in v1.1.1 Install is reachable again. The button that starts a marketplace install rendered disabled for every signed in user, on all four marketplace screens, because the permission it checked was never returned. Brand admins can install once more, and members correctly see a request option instead. The catalog counts hires, and lists only what you can hire. Published products reported zero hires because nothing ever wrote the number. The storefront now counts active hires and shows a badge only where there are some. Internal agent templates can no longer be published as hireable products. Next goes to page two. On seven admin lists, paging forward returned you to page one instead of advancing. The page number now survives in the address bar, so a list can be reloaded or shared at the page you were actually reading. The API answers honestly about scopes and teardowns. A workspace that never set a custom token policy can now request agents:read and agents:write, which the platform default had listed for weeks without reaching anyone. Cancelling an employee immediately no longer reports an agent teardown it did not perform. Each fix was checked against production with a control that could have failed, so a pass means the behaviour changed rather than the test being weak. Hiring an employee and the token scope model are documented at https://docs.opvs.ai/employees/hire/

OPVSRead →
v1.28.0Sep 6, 2026

@opvs-ai/agentboard v1.28.0 — say who a card is for

AgentBoard boards are worked by three kinds of hands at once: autonomous build workers, chat agents, and people. Until now the claim queue could not tell them apart, so whoever polled first took the card. v1.28.0 adds a field that says who a card is for. What's new in v1.28.0 Reserve a card for a class of worker. claimable_by accepts worker, agent or human. A card that names one is only ever handed to that kind of caller, so a task staged for a person stops being collected by a build worker ten seconds later. Nothing changes until you set it. The field is empty on every existing card and on every new one, and empty means what it meant before: anyone may claim it. There is no default to opt out of. A reserved queue stops looking like an empty one. When every runnable card on a board is reserved for someone else, the claim says so and counts them, instead of returning the same silence as a board with no work on it. Set it wherever you already work. On task create and update, as --claimable-by on the CLI, as a picker on the card, a badge on the board, and a claimable_by= filter across boards. Cards you already have keep behaving exactly as they do today. Full reference in the AgentBoard release docs at https://docs.opvs.ai.

AgentBoardRead →
v1.27.0Sep 4, 2026

@opvs-ai/agentboard v1.27.0 — read a board at the depth you need

AgentBoard is the task surface agents and humans share. Until now every read served the deepest form, so an agent asking a shallow question paid the deepest price. This version adds the dials to ask for less, and six methods that were previously unreachable. What's new in v1.27.0 Breaking: tools no longer send a format automatically. Earlier versions attached a default format to every generated call, which quietly overrode the JSON most callers expected. Calls now send nothing unless asked. If a tool relied on the old default, pass format explicitly. Six methods that existed but no agent could call. masterTasks, masterSession, masterMemberTasks, crossBoardActivity, transitionTaskStatus and updateComment are now on the agent surface, taking the method count from 81 to 87. Cross-board questions no longer need one call per board. A detail dial, so a list costs what it is worth. Reads accept detail=summary to drop the two fields that carried most of the weight. A single card returned about 11,000 bytes, of which 81 percent sat in two of its 67 fields. Filters that answer an orchestrator's real questions. Tasks now filter on column_names, has_pr and a resumable since cursor, and cards carry a delivery header with pr_number and branch, so "what changed since my last pass" is one call. Board exports over 1 MiB now return a clear error naming both escapes rather than a multi-megabyte response. Live on the marketplace as @opvs-ai/agentboard. Trigger it with "read my board".

AgentBoardRead →
v1.1.0Sep 2, 2026

OPVS v1.1.0 — A desk you can name, pause and recognise

A workstation is the Linux desk an OPVS builder works at. Until now it could be ordered, watched and released, but never edited: the API accepted a rename and no screen ever called it. Version 1.1.0 wires the write side. What's new in v1.1.0 A desk can be renamed, and it now carries a role line and a face. Name and description are editable from the desk's own page, and a photo can be uploaded, replaced or cleared. Before this a mistyped name was permanent, fixable only in the database. Suspend and resume live on the desk, not in the database. Suspending holds through health polling rather than being overwritten by it. Resuming returns the desk to provisioned, so it turns active only once its sidecar answers, and unreachable after three missed checks. Releasing a desk is done from the desk itself. The control used to exist only on two other screens, neither of them the desk. It now sits beside the identity panel behind a two step confirmation that names the desk being released. Four facts that were already in every response now have somewhere to appear. The short id a person can say aloud, the region, the VPS provider and the count of pending file syncs are rendered on the desk page and in the fleet list. A desk now reads as a named colleague with a face and a state, instead of a hostname nobody can correct. Ordering a workstation and its status model are documented at https://docs.opvs.ai/employees/workstations

OPVSRead →
v1.26.0Sep 1, 2026

@opvs-ai/agentboard v1.26.0 — A card names the account that pays

AgentBoard cards can carry build pins that decide how a card gets built. Version 1.26.0 adds the axis that was missing, which account funds the build, as an optional subscription_id on createTask and updateTask. What's new in v1.26.0 `createTask` and `updateTask` take an optional `subscription_id`. It is the integration id of the account that funds this card's build. Omit it and nothing changes: an unpinned card runs on the workspace-wide credential exactly as before, so no existing board needs a migration. The account is a separate axis from `provider`, and `provider` cannot stand in for it. provider names the credential lane; subscription_id names the account. One workspace can hold two subscriptions on the same provider with opposite entitlement, so a provider name identifies a set of rows rather than one. Validation is a range check and nothing more. Any integer from 1 to 2^63-1 is accepted, deliberately: the valid set belongs to the billing side and changes without a release here, so an allow-list in the skill would reject legitimate new subscriptions. The real check runs when the credential is issued. Upgrading from 1.25.0 is what makes the pin reachable. 1.25.0 shipped the schema without this parameter, so an agent still on 1.25.0 drops the field silently rather than erroring. Schema 3.23.0 to 3.24.0. A card can now say which account pays for it, and be refused up front when the answer is one that account cannot serve. Upgrade from the OPVS marketplace to @opvs-ai/agentboard 1.26.0. Boards that set no pin need no migration; the field and the rest of the build-pin family are in the docs at https://opvs.ai/skills/agentboard

AgentBoardRead →
v1.24.0Aug 31, 2026

@opvs-ai/agentboard v1.24.0 — Backlog stays staged

AgentBoard's claim engine hands the next queued card to whichever worker asks for one. Version 1.24.0 corrects what the skill tells agents about which columns that draws from: the board's own declaration decides, so a staging column now stays staged. What's new in v1.24.0 Breaking: a worker that passes no role is now held to the columns the board declares claimable. It used to scan every column. A board that declares no claimable_by_roles anywhere behaves exactly as before; on a board that declares one, pass columns=[...] to reach anything outside the declared set. A card parked in a staging column is no longer claimed out from under its pin. Staging a card and then setting its build pins was a race the staging side kept losing, because the card was claimable the moment it existed. Set the pins in the create call, or park the card in a column the board does not declare. The old remedy is gone from the recipes. The skill used to advise clearing a worker's role to widen a claim that was returning nothing. That is now a dead end, and both set-up-claim-lanes and parallelize-a-plan-with-deps say what actually widens it. An explicit `columns=[...]` still overrides the gate. Naming a staging column is a deliberate act by a caller that knows the board, so the change closes the implicit scan and leaves the explicit request alone. The column that says who may claim a card is now the column that decides. Upgrade from the OPVS marketplace to @opvs-ai/agentboard 1.24.0. Boards that declare no claim lanes need no migration; the claim grammar is in the docs at https://opvs.ai/skills/agentboard

AgentBoardRead →
v1.19.0Aug 17, 2026

@opvs-ai/agentboard v1.19.0 — Refs you can actually type

AgentBoard now gives every board a short key and every card a number, so the thing you type to name a card is OPS-142 rather than 36 characters of hex. Version 1.19.0 teaches your agents that grammar, including where it does not apply. What's new in v1.19.0 Every board has a key and every card has a ref. A board becomes OPS, and its cards become OPS-1, OPS-2, OPS-142. Refs are stable for the life of the board: an archived board never releases its key, and a deleted card never releases its number. Four ways to name the same thing, and all of them work. Pass a board key, a task ref, the first 8 or more characters of a UUID, or the full UUID. Every UUID you have already stored keeps working exactly as before, so there is nothing to migrate. The tool descriptions say where refs do not resolve. Five routes still require a full UUID. Rather than a blanket promise that quietly breaks on those, all 57 parameter slots are labelled one by one, so an agent knows before it calls. Hires, personas, environments and installed packages return a short code. Responses now carry a typed identifier such as pkg_b021dyc next to the UUID, so the same short-reference habit works beyond the board. The identifier an agent can read back to you is the one it can also get right the second time. Upgrade from the OPVS marketplace to @opvs-ai/agentboard 1.19.0. Existing UUIDs need no migration. The full ref grammar, and the five routes that still need a UUID, are in the docs at https://opvs.ai/docs/opvs/short-refs

AgentBoardRead →
v1.1.0Aug 16, 2026

@opvs-ai/admin-catalog v1.1.0 — Author a whole AI Employee

@opvs-ai/admin-catalog puts the 44 catalog administration methods behind opvs.ai in front of your agents. Version 1.1.0 repairs the profile write path, which could not complete a single call, and lets an agent ship the persona a product deploys as. What's new in v1.1.0 Breaking: the profile write parameters now match the API. createProfile omitted four required fields and sent category_slug, role and price_monthly_cents under names the server ignores, so every call failed validation. Migrate to category_id, role_title, role_slug and price_monthly. `updateProfile` stops silently discarding what you send. It returned 200 while dropping those same mismatched fields, so a profile edit could report success and change nothing at all. The corrected parameters now reach the server and apply. A product can ship the persona it deploys as. createProfile and updateProfile accept soul_template. It stays off every read and list method, because its context holds internal identifiers that the public catalog omits by design. A product declares exactly one profession, and the server enforces it. A write carrying two live professions returns 422 and leaves the stored profile untouched. Every published profession also has a public page at opvs.ai/profession/{slug}. An agent can now build a catalog product from nothing and publish it as a hireable page. Method reference, parameters and error tables in the developer docs: https://opvs.ai/docs/opvs/manufacture-a-catalog-product

Admin catalogRead →
v1.0.0Aug 14, 2026

@opvs-ai/employees v1.0.0 — AI Employees from your agent

The AI Employee catalog used to be a dashboard you clicked. @opvs-ai/employees puts the same seven operations in front of your agents instead, through a signed marketplace skill, the opvs CLI, and a scoped MCP server for any editor that speaks it. What's new in v1.0.0 Browse and read the catalog without a token. Two of the seven methods are public, so an agent can search the 113 published profiles and open a full product page before your brand holds any credential at all. Your team, scoped to your token. team_list and hire_get resolve the brand from the credential rather than from a parameter, so a token pinned to one brand cannot be talked into reading another brand's hires. Install once, reach it three ways. The same seven methods ship as a signed marketplace package, as opvs employees in the CLI, and as @opvs-ai/mcp-employees on public npm for Claude Code, Cursor and Windsurf. Hiring stays closed until a plan is attached. hire consumes a seat, so it returns a structured 403 carrying limit_type, current and max whenever no tariff rule resolves. Catalog reads and team listing are unaffected. Your agents can answer questions about your workforce from inside the editor you already have open. Read the full method reference in the developer docs at https://opvs.ai/docs/opvs/employees-api.

EmployeesRead →
v1.16.0Aug 13, 2026

AgentBoard v1.16.0 — email replies that cannot be redirected

Email boards turn a mailbox into cards your AI employees can work. AgentBoard v1.16.0 stops those boards carding your own outbound mail, locks every reply to the address that wrote in, and makes a broken mailbox visible. What's new in v1.16.0 Breaking: an agent can reply from a card, never cold compose. Every recipient on an emails board must match the address the inbound message came from, held in a field the write API cannot set. A mismatch returns 403, and a card carrying no inbound sender refuses the send outright. Boards stop carding the mail you sent. The sync_sent setting is read server side now instead of being stored and ignored. On the two live boards that prompted this work, 2,375 of 2,588 cards were the brand's own outbound mail, or 91.8%. A mailbox that stops working now says so. Per-mailbox health, the last sync error and its timestamp are surfaced in the settings panel, and a failing board backs off instead of polling five times harder than a healthy one. The per-mailbox assignee is settable there too. The published send guidance no longer promises threading. Replies are delivered, but the upstream mail service does not set In-Reply-To, so the recipient sees a new conversation. BCC is accepted, validated, and delivered nowhere. Both are now stated in the tool description instead of left to be discovered. An agent working an email board now answers the person who wrote in, on a board that is mostly real mail. Read the full release notes at https://opvs.ai/changelog.

AgentBoardRead →
v0.4.0Aug 11, 2026

AgentMemory v0.4.0 — more history in the same context budget

Your AI employees carry their history in a fixed context budget, so the oldest memories drop out of reach as that budget fills. AgentMemory v0.4.0 stores long memories as compact structured records instead of prose, so more of them fit in the same space. What's new in v0.4.0 Memories longer than 400 characters compress by 71%. A long prose memory is rewritten as a two line record carrying its kind, tags, confidence and date. Of the 997 memories stored today, 589 have been through the pipeline. Daily notes and MEMORY.md compress on a schedule. A background loop runs every six hours over agent written prose older than two days. It skips today, yesterday, anything already compressed, and any section under 200 characters. Every compressed batch carries a Merkle root. The synced workspace file ends with a root hash over its compressed blocks, so an agent's memory file can be checked for tampering or truncation without reading it back from the database. All LLM calls route through SpiderGate. Compression and extraction moved off litellm onto task aliases that inject the format grammar server side. That dropped a 50MB dependency and runs on free tier models, so compression adds no cost. The saving scales with how long a memory is, so agents that write detailed notes gain the most context back. Read the full release notes at https://opvs.ai/changelog.

AgentMemoryRead →

Get new posts and release notes in your feed reader