Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx skills add GeminiLight/MindOS --skill mindos-viewsgit clone --depth 1 https://github.com/GeminiLight/MindOSWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/skills/geminilight/mindos/mindos-views)<a href="https://agentmods.dev/skills/geminilight/mindos/mindos-views"><img src="https://agentmods.dev/badge/skills/geminilight/mindos/mindos-views/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/skills/geminilight/mindos/mindos-views"><img src="https://agentmods.dev/badge/skills/geminilight/mindos/mindos-views.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00060 | $0.01697 |
| Opus 5.5 | $0.00024 | $0.00679 |
| Sonnet 5.5 | $0.00012 | $0.00339 |
| Haiku 4.5 | $0.00006 | $0.00170 |
Grade A, and why
mindos-views scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured today.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 55 lines — stays where its author put it; the contents beside it link to each section on GitHub.
MindOS shared work views
Use the mindos_views_* tools for records/single-source views and mindos_pages_* for multi-source work pages. If MCP is unavailable, use mindos views with a JSON file or stdin. Never bypass this workflow by editing the registry, CSV, or Markdown directly. The same runtime, root and original files are shared by Codex, Claude Code and the human UI.
- Call
mindos_views_describeand requirecapabilityVersion: 1. If unavailable, run the packaged setup skill. Never installlateston each session or create a second library to fix a connection. - Read the existing views and collections first. Reuse a matching collection. Creation requires a user-requested folder and an operation UUID; choose
peopleortasks, andneworexisting. Do not automatically migrate arbitrary CSV or TODO.md. - Read current records and schema before editing. Keep
rootId, record source revision, schema revision, collection ID, record ID and a fresh UUID for each intended mutation. Never use row position as identity. Forward all expected revisions. For task creation, include the collection file revision; for a new person, omit the content revision. - Use only supported actions:
record.create,record.update,interaction.add,followup.update,schema.extend,operation.undo. For optional task text fields use a newx_key. Explain the new field and empty initial values before the user-authorized change. Preserve unknown/private content; structured read is intentionally a field projection, not all source text. - To generate a UI, use the current field catalogue and construct a small ViewSpec:
specVersion: 1,collectionId,layout,fields, optionalsort,filter.all,labels, andactions. Layouts:list-detail,table,board; board groups taskstatus. For a ViewSpec, never generate HTML, React, JavaScript, CSS, SQL, URLs or executable actions; custom HTML belongs only in a PageSpec html block below. Labels are text. Dates use YYYY-MM-DD. Filters have at most 10 conditions and sort at most 2 fields. - Example requests: “按状态分组显示项目待办,隐藏已完成” → board/groupBy status/filter neq done; “按姓名排序的人脉表格” → table/name, background, focus/sort name asc. Call preview first and show its URL. Save if the user already authorized the adjustment; otherwise let them inspect the preview. Use expectedViewVersion when replacing a saved view.
copy,restore,removechange views only; removing never deletes original records. - Read back after saving and report the actual source path, view URL and committed result. URLs returned by tools are relative to the connected MindOS origin;
mindos views open <view-id>validates and opens them. Never put bearer tokens into URLs. - On conflict, preserve the intended edit, re-read and compare; do not silently retry with a fresh revision. On timeout, query
mindos_views_readwith the SAME operationId; retry the IDENTICAL request. Never change input under the same ID. A successful receipt confirms that operation, not that later edits are absent. Undo updates only while the source still matches the saved after revision; creation cannot be undone.
CLI examples (JSON output):
mindos views doctor
mindos views list
mindos views describe c-collection-id
mindos views read c-collection-id
mindos views create --input request.json
mindos views mutate --input request.json
mindos views preview --input request.json
mindos views save --input request.json
mindos views operation operation-uuid
mindos views open v-view-id
All write JSON includes expectedRootId; mutation/create/save includes operationId. A server failure is not permission to write files directly. The plugin adds no model account and does not send messages or publish content. It cannot isolate a malicious local agent that already has filesystem access.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- today Changed · +20 lines · +12 tokens per session aab63a66d6e6
- yesterday First seen · 35 lines · 48 tokens per session scan A 5b106419e8ba
mindos-views is a skill published in the GitHub repository GeminiLight/MindOS (680 stars, last pushed today), licensed MIT. It adds 60 tokens to every session and 1,697 once invoked, about $0.0002 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-10-22.
Other skills, from other repositories
core
Use when knowledge base hub — PARA-structured company memory combining company-kb and kb for persistent context, project documentation, and agent recall across sessions. Use when working with knowledge base, company knowledge, or persistent memory.
kb-memory
Use when knowledge base and memory system for AI agents. Covers company KB, persistent memory, session recall, and brain architecture for context preservation.
acontext-installer
Install Acontext, Login & Init Acontext Project, Add Skill Memory to Agent.
babysit
Use when a user asks to babysit, monitor, keep checking, keep an eye on, or report when a pull request, CI run, ticket, deployment, or other changing target reaches an outcome.
openlore
Query and publish to an OpenLore knowledge repository over SSH using ordinary shell commands. Use when a task needs project documentation, runbooks, shared team knowledge, or a place to publish findings.
openlore-housekeeping
Audit and maintain a shared OpenLore knowledge repository. Use on a schedule or on request to find stale docs, broken links, unreviewed inbox items, and missing skill coverage, then publish an audit report.