extensions
Skill0 starsChanged 5 months agoMITView on GitHub
Build, modify, explain, validate, package, or install Mantis extensions and custom vertical panels. Use when a user asks for reusable Mantis UI, extension commands or events, a .mantisx package, or an optional Python-backed extension action.
2 other projects use a version of this.
Adopted by
Most recently taken up by
- BuilderIO/agent-nativetheir own version, first seen Oct 8, 2026
- KellisLab/mantis-cliword for word, first seen Jun 27, 2026
Other versions2 projects use a version of this.
Word for word · 1
- KellisLab/mantis-cli★ 2100% same
Edited · 0
None found.
Wrote their own version · 1
- BuilderIO/agent-native★ 7.6ksame name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
The file
Show its rules and text
We split this skill into the rules it gives an agent
21 rules. About 0 are common, 15 are distinctive, and 6 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈15
Distinctive · estimate
This author's own choices. Worth a look.
6
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 18
Run this before any other Mantis command or tool, unless it already ran this turn:
- Distinctive: line 24
Use the returned space id when installation needs a target.
- Distinctive: line 48
use panel-local code for UI and temporary view state
- Distinctive: line 49
add a host main only for durable commands, subscriptions, activation, or state
- Distinctive: line 50
add Python only when the browser SDK cannot do the required work
- Distinctive: line 54
Run the project's tests/build when present, package it, then inspect the archive layout before reporting success.
- Very specific: line 58
Every package needs mantis.extension.json.
- Distinctive: line 59
Keep extension, panel, and command ids stable.
- Very specific: line 61
Use context.workspaceState / window.mantis.workspaceState for state scoped to the current space and globalState for user-level extension …
- Very specific: line 62
Put durable command registrations and event subscriptions in host activate(context) and add returned disposables to context.subscriptions.
- Very specific: line 64
Do not use raw rpc unless debugging the bridge and the documented helper cannot serve the request.
- Very specific: line 65
Do not use unsupported contributes.menus or contributes.settings fields.
- Distinctive: line 71
When a backend is needed:
- Very specific: line 77
keep network: false unless external access is essential and explicitly disclosed
- Distinctive: line 78
never treat the backend context as secret storage
- Distinctive: line 80
Do not describe the backend as deprecated unless the current docs say so.
- Distinctive: line 84
Treat extensions as trusted code even though panels are sandboxed and SDK calls are permission-gated.
- Distinctive: line 85
Never add same-origin access to a panel or reach into the parent DOM, cookies, or browser storage.
- Distinctive: line 86
Do not embed API keys or other secrets in browser or backend assets.
- Distinctive: line 87
Avoid sending Mantis data to external services.
- Distinctive: line 88
Prefer explicit user actions for writes, clear progress for slow work, and visible errors for permission or backend failures.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Mantis Extensions
Mantis extensions are trusted packages that add reusable workspace functionality without changing the core app. They can contain:
- sandboxed panel UIs, available under Verticals > Extensions
- an extension host for durable commands, subscriptions, activation, and state
- an optional Python backend for work that should not run in the browser
Panel code uses window.mantis; host code uses mantis. Both use the same asynchronous, permission-gated SDK. Panels are short-lived and run in allow-scripts iframes without same-origin access. The host runs in a Web Worker and survives panel closure until the extension deactivates.
First step
Run this before any other Mantis command or tool, unless it already ran this turn:
mantis use get_space_contextUse the returned space id when installation needs a target. Never guess a space, map, field, or permission.
Documentation routing
The docs are authoritative. Read the pages relevant to the requested contribution; do not invent manifest fields, SDK methods, permissions, package layouts, or lifecycle behavior.
| Need | Read |
| Architecture or choosing host vs panel | Overview |
| Any new or changed package | Manifest |
| Panel layout and lifecycle | Custom panels and verticals and Panel UI |
| Mantis data, state, commands, or events | SDK API |
| Server-side execution | Python backend |
| Producing or checking an archive | Packaging and installation |
| A concrete SDK pattern | Examples |
| Permissions, trust, or platform limits | Security and limits |
For a normal panel extension, read the overview, manifest, relevant panel/SDK sections, packaging, and security guidance. Read backend documentation only when server-side execution is actually needed. Use examples as patterns, not as a substitute for the reference pages.
Build workflow
- Clarify the user-visible workflow, what Mantis data it reads or writes, whether behavior must survive panel closure, and the target space if installation is requested.
- Inspect the existing extension project before choosing its build system or layout. Preserve established conventions.
- Design the smallest useful contribution:
- use panel-local code for UI and temporary view state
- add a host
mainonly for durable commands, subscriptions, activation, or state - add Python only when the browser SDK cannot do the required work
- Declare the narrowest supported permission set. Explain every write, command, backend, or network capability.
- Build self-contained browser output. Do not import from the parent Mantis app; bundle host and panel dependencies into the extension output.
- Validate manifest paths and ids, contribution entries, activation events, permissions, backend actions, and referenced assets.
- Run the project's tests/build when present, package it, then inspect the archive layout before reporting success.
Manifest and SDK rules
- Every package needs
mantis.extension.json. - Keep extension, panel, and command ids stable. Use globally unique dotted ids for extensions and prefix command ids with the extension id.
- Declare only documented permissions:
maps:read,selection:read,selection:write,bags:write,panels:write,commands:execute, andbackend:invoke. - Use
context.workspaceState/window.mantis.workspaceStatefor state scoped to the current space andglobalStatefor user-level extension state. - Put durable command registrations and event subscriptions in host
activate(context)and add returned disposables tocontext.subscriptions. - Dispose panel-local subscriptions when the panel unloads.
- Do not use raw
rpcunless debugging the bridge and the documented helper cannot serve the request. - Do not use unsupported
contributes.menusorcontributes.settingsfields.
Python backend decision
Python backends are supported but optional. Add one only for server-side computation, Python-only dependencies, or work that cannot safely or practically run in the panel/host SDK.
When a backend is needed:
- request
backend:invoke - allowlist actions in
backend.actions - validate every payload and return JSON-serializable values
- pin and minimize requirements
- keep
network: falseunless external access is essential and explicitly disclosed - never treat the backend context as secret storage
Do not describe the backend as deprecated unless the current docs say so.
Security
- Treat extensions as trusted code even though panels are sandboxed and SDK calls are permission-gated.
- Never add same-origin access to a panel or reach into the parent DOM, cookies, or browser storage.
- Do not embed API keys or other secrets in browser or backend assets.
- Avoid sending Mantis data to external services; if required, make the data flow and backend network access explicit.
- Prefer explicit user actions for writes, clear progress for slow work, and visible errors for permission or backend failures.
- Review publisher/source, permissions, dependencies, backend actions, and network access before installation.
Package and install
Prefer .mantisx or .zip for real extensions; use inline JSON only for small fixtures or debugging. The archive must contain mantis.extension.json at its root or inside one single top-level folder. Browser assets must not be under backend/; Python backend files must be under backend/. Paths must be safe, relative, and UTF-8 text.
Installing changes a Mantis space. Install only when the user asks for it or the task explicitly includes installation:
mantis use install_extension --file extension.mantisx --space-id <space-id>The API-key user must own the target space. Installation is personal to that user within the space. Report the extension id/version, panels and commands, requested permissions, backend/network status, archive path, and installation result.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

