feat: initial commit - Jhonny Editor
- Adicionado estrutura completa do projeto - Configurado MCP server para Premiere Pro - Adicionado documentação e skills - Configurado Gitignore para o projeto
This commit is contained in:
João Henrique
+69
@@ -0,0 +1,69 @@
|
||||
# Adobe Video Partner brief
|
||||
|
||||
**Status:** prepared for an owner-operated Adobe Video Partner application.
|
||||
This document is not an application, endorsement, Marketplace listing, or
|
||||
claim of Adobe affiliation.
|
||||
|
||||
## Suggested application summary
|
||||
|
||||
MCP for Adobe Premiere Pro is a free, MIT-licensed local integration that lets
|
||||
compatible AI clients use structured, capability-aware Premiere workflows. It
|
||||
is designed for reviewable work: start with a read-only connection check,
|
||||
inspect supported project state, preview meaningful changes where available,
|
||||
and return explicit results or limitations. The recommended setup keeps the AI
|
||||
client, server, connector, and Premiere host on the editor's computer.
|
||||
|
||||
The project complements Adobe's evolving native AI experiences. It does not
|
||||
claim to replace them, to be affiliated with Adobe, or to provide unattended
|
||||
editing, universal host compatibility, or a hosted relay into a customer's
|
||||
desktop Premiere process.
|
||||
|
||||
## Links for the application owner
|
||||
|
||||
- Adobe Video Partner Program: <https://www.adobevideopartner.com/>
|
||||
- Product site: <https://premiere-pro-mcp.com/>
|
||||
- Source and support: <https://github.com/leancoderkavy/premiere-pro-mcp>
|
||||
- Current release: <https://github.com/leancoderkavy/premiere-pro-mcp/releases/latest>
|
||||
- [Installation and local-first boundary](../../README.md)
|
||||
- [Capability and support boundary](../supported-actions.md)
|
||||
- [Marketplace release checklist](../adobe-marketplace-release-checklist.md)
|
||||
- [Design-partner pilot controls](../industry/security-and-design-partner-pilot.md)
|
||||
|
||||
## Evidence the owner should attach or link
|
||||
|
||||
1. A current release build and the Marketplace-targeted connector package only
|
||||
when it has passed the channel-specific validation.
|
||||
2. Three real, redacted licensed-host walkthroughs: read-only connection
|
||||
verification, a review-first workflow, and a returned verification or
|
||||
limitation. Label simulations and illustrations as such.
|
||||
3. The exact host version, operating system, artifact hash, workflow boundary,
|
||||
and observed result for every claimed demonstration.
|
||||
4. Current privacy, security, and support pages, plus the reviewer instructions
|
||||
needed to install and test the local connector.
|
||||
|
||||
## Conversation goals
|
||||
|
||||
- Ask for feedback on UXP integration and distribution expectations.
|
||||
- Offer a narrowly scoped design-partner evaluation using synthetic or
|
||||
facility-approved fixtures, not production media by default.
|
||||
- Invite review of one repeatable workflow—Project Intake, review planning, or
|
||||
delivery preflight—rather than claiming general autonomous editing.
|
||||
|
||||
## Do not say
|
||||
|
||||
- "Adobe-approved," "Adobe-endorsed," or "Adobe-certified" before Adobe
|
||||
independently grants that status.
|
||||
- "Marketplace available" before a public Marketplace URL is live.
|
||||
- "Production-proven," "safe for every project," or "works on every Premiere
|
||||
version" without evidence for the specific workflow and host.
|
||||
- That local-first architecture changes the privacy terms of a chosen AI client
|
||||
or model provider.
|
||||
|
||||
## Owner actions still required
|
||||
|
||||
1. Use the Adobe Video Partner Program's current application path and provide
|
||||
the final publisher/contact details.
|
||||
2. Verify current Marketplace reviewer requirements in Adobe Developer
|
||||
Distribution; account status and requested materials are time-sensitive.
|
||||
3. Approve any public contact, customer reference, screenshot, or demonstration
|
||||
before it is sent or posted.
|
||||
Executable
+73
@@ -0,0 +1,73 @@
|
||||
# Premiere Pro MCP community launch kit
|
||||
|
||||
**Status:** prepared drafts only. Nothing in this file has been posted, sent, or scheduled.
|
||||
|
||||
## Use this only when it helps the current conversation
|
||||
|
||||
- Check each community's self-promotion rules before posting.
|
||||
- Answer the workflow question first. Link a guide only when it genuinely answers the follow-up.
|
||||
- Never represent illustrated site assets as a live Premiere session.
|
||||
- Do not claim every tool or host version will work. Direct readers to the read-only connection check and capability inspection.
|
||||
- Do not say the local server changes the privacy terms of the AI client the reader chooses.
|
||||
- Do not contact people from this document without an approved contact list and owner authorization.
|
||||
|
||||
## Owned social draft
|
||||
|
||||
AI editing is most useful when it removes repeated Premiere work without hiding the edit.
|
||||
|
||||
Premiere Pro MCP is a free, MIT-licensed local bridge between a compatible AI client and supported Premiere workflows. Start with a read-only connection check, inspect the project, preview a bounded request, and verify the returned result before relying on it.
|
||||
|
||||
Practical setup and workflow guides: `https://premiere-pro-mcp.com/blog/?utm_source=linkedin&utm_medium=organic_social&utm_campaign=premiere_workflow_guides`
|
||||
|
||||
## Technical-community draft
|
||||
|
||||
I maintain an open-source MCP server for supported Adobe Premiere Pro workflows. The goal is not autonomous editing. It is to make repeated work inspectable: check the connection, inspect a project, create a non-mutating plan where available, preview it, and verify the returned result.
|
||||
|
||||
The project is local-first and independent from Adobe's AI Assistant. This guide compares the workflows without claiming either is universally better:
|
||||
|
||||
`https://premiere-pro-mcp.com/blog/adobe-premiere-ai-assistant-vs-mcp/?utm_source=community&utm_medium=organic_referral&utm_campaign=premiere_workflow_comparison`
|
||||
|
||||
Useful feedback: installation friction, capability boundaries, and the first repeatable Premiere task a team would test.
|
||||
|
||||
## Editorial-community reply pattern
|
||||
|
||||
1. Name the specific Premiere task the person is trying to repeat.
|
||||
2. Share one concrete, non-promotional way to make it safer: define the target sequence, what must not change, and how the result will be checked.
|
||||
3. If the person asks for a tool or implementation, disclose the project relationship and link the one relevant guide.
|
||||
4. Invite correction or workflow details. Do not push a download.
|
||||
|
||||
## Design-partner interview invitation draft
|
||||
|
||||
**Subject:** Could we map one repeatable Premiere workflow?
|
||||
|
||||
Hi [name],
|
||||
|
||||
I maintain Premiere Pro MCP, an open-source local bridge for reviewable Premiere workflows through compatible AI clients. I am looking to learn how post teams handle one repeated task such as project intake, cutdown preparation, or delivery checks.
|
||||
|
||||
This is a 30-minute research conversation, not a product-sale call. We will map the current steps, what must stay under editor control, and what evidence would make an automation trustworthy. We will not ask for project media, client footage, credentials, or confidential project names.
|
||||
|
||||
If it is useful, I can send the workflow questions in advance.
|
||||
|
||||
Thanks,
|
||||
Premiere Pro MCP contributors
|
||||
|
||||
## Interview guide
|
||||
|
||||
- Which Premiere task repeats often enough to document?
|
||||
- What triggers it, who owns it, and where does handoff fail?
|
||||
- What input must an assistant see, and what must it never change?
|
||||
- What output would prove the workflow succeeded?
|
||||
- Which host versions and AI clients are involved?
|
||||
- What would make setup too risky or too time-consuming?
|
||||
- May we retain this feedback anonymously? Do not record customer claims or quotes without separate written approval.
|
||||
|
||||
## Measurement
|
||||
|
||||
Use campaign links with the existing allowlisted UTM format:
|
||||
|
||||
- `utm_source`: channel or named partner
|
||||
- `utm_medium`: `organic_social`, `referral`, or `email`
|
||||
- `utm_campaign`: `premiere_workflow_guides` or another stable lowercase campaign ID
|
||||
- `utm_content`: a stable creative ID when variants are tested
|
||||
|
||||
Primary success signal: a completed read-only connection check and an observable first workflow result. Likes, impressions, stars, and downloads are diagnostic only.
|
||||
+42
@@ -0,0 +1,42 @@
|
||||
# MCP Registry readiness
|
||||
|
||||
**Status:** v1.14.8 local and published-npm metadata verified on 2026-09-04; not submitted to the official registry.
|
||||
|
||||
The official MCP Registry is a separate public listing. A repository change, an npm package, or a GitHub release does not create a listing. Submission requires a user-authorized registry login and publication action.
|
||||
|
||||
The verified v1.14.8 npm artifact exposes
|
||||
`mcpName: io.github.leancoderkavy/premiere-pro`, and its checked-in `registry/server.json`
|
||||
matches the published package name, version, repository, and local `stdio`
|
||||
transport. Run the official validator and the read-only preflight below at the
|
||||
moment of submission; their results are readiness evidence, not a publication
|
||||
claim.
|
||||
|
||||
Before an owner-authorized submission:
|
||||
|
||||
1. Run `npm run validate:mcp-registry-metadata` followed by
|
||||
`npm run preflight:mcp-registry`. The latter checks the published npm
|
||||
artifact and searches the official registry; it does not authenticate or
|
||||
publish.
|
||||
2. Inspect `registry/server.json` from the exact release tag and confirm the
|
||||
entry continues to describe only the local `stdio` route. Do not present the
|
||||
local Premiere bridge as a hosted service.
|
||||
3. Obtain action-time approval, authenticate with `mcp-publisher login github`,
|
||||
and publish once with `mcp-publisher publish registry/server.json`.
|
||||
4. Query the registry for the exact listing name and retain the returned URL as
|
||||
public evidence. Do not create a second record merely because search results
|
||||
are delayed.
|
||||
|
||||
Use only these evidence-bounded facts in a future directory entry:
|
||||
|
||||
- Free, MIT-licensed, local-first MCP server for supported Premiere Pro workflows.
|
||||
- A compatible AI client calls structured tools through a local Premiere connection.
|
||||
- Begin with `verify_premiere_connection`; support remains capability- and host-dependent.
|
||||
- The chosen AI client's privacy behavior is separate from the server's local-first recommendation.
|
||||
|
||||
Do not submit until the package version, transport, authentication expectations, privacy disclosures, and source URL are all verified for that directory.
|
||||
|
||||
Official references:
|
||||
|
||||
- <https://modelcontextprotocol.io/registry/quickstart>
|
||||
- <https://registry.modelcontextprotocol.io/docs>
|
||||
- <https://modelcontextprotocol.io/registry/faq>
|
||||
Executable
+98
@@ -0,0 +1,98 @@
|
||||
# MCP for Adobe Premiere Pro Organic Distribution Kit
|
||||
|
||||
## Gate before any public distribution
|
||||
|
||||
Treat this file as draft copy, not proof that a guide is public. Before a post,
|
||||
outreach message, directory submission, or community link goes out, verify the
|
||||
exact production guide URL and `/blog/` return HTTPS 200 on the intended
|
||||
deployed commit. Do not publish a URL from a local build or repository branch.
|
||||
|
||||
## Guide inventory
|
||||
|
||||
| Reader intent | Guide | Public route to verify before sharing |
|
||||
| --- | --- | --- |
|
||||
| Understand the product | What is an MCP server for Adobe Premiere Pro? | `/blog/what-is-a-premiere-pro-mcp-server/` |
|
||||
| Evaluate a safe first workflow | Premiere Pro AI workflow checklist | `/blog/premiere-pro-ai-workflow-checklist/` |
|
||||
| Compare AI routes | Adobe Premiere AI Assistant vs. MCP | `/blog/adobe-premiere-ai-assistant-vs-mcp/` |
|
||||
| Prepare assistant-editor intake | Premiere Pro Project Intake checklist | `/blog/premiere-pro-project-intake-checklist/` |
|
||||
| Preserve a recovery point | Premiere Pro project backup checklist | `/blog/premiere-pro-project-backup-checklist/` |
|
||||
| Focus a visual review | Premiere Pro review frames and scene detection | `/blog/premiere-pro-review-frames-and-scene-detection/` |
|
||||
| Inspect a delivery file | Premiere Pro delivery QC and loudness checklist | `/blog/premiere-pro-delivery-qc-and-loudness-checklist/` |
|
||||
|
||||
## Positioning
|
||||
|
||||
MCP for Adobe Premiere Pro gives compatible AI assistants a structured, local-first way to inspect Premiere projects, plan supported edits, automate repeatable work, and return observable results—without replacing the editor’s creative decision.
|
||||
|
||||
## Draft release post
|
||||
|
||||
When the checked guide URLs are live, share this for editors and workflow teams who want to use AI with Adobe Premiere Pro without handing over creative control:
|
||||
|
||||
1. A practical comparison of Adobe’s current AI Assistant beta and a structured MCP workflow
|
||||
2. A project-backup checklist before high-risk automation or organization work
|
||||
3. A visual-review guide for file-verified frames and source scene candidates
|
||||
4. A delivery-QC guide for scoped black/freeze findings and loudness measurement
|
||||
|
||||
MCP for Adobe Premiere Pro is free, MIT licensed, and local-first. Start with a read-only connection check, then inspect your active sequence before requesting an edit.
|
||||
|
||||
Read the guides: https://premiere-pro-mcp.com/blog/
|
||||
|
||||
## Short social variants
|
||||
|
||||
### LinkedIn
|
||||
|
||||
AI video editing should make repetitive Premiere work easier to review—not make creative decisions in a black box.
|
||||
|
||||
When the guide set is live, share the comparison, project-backup, visual-review, and delivery-QC checklists with the post-production teams who need a bounded way to evaluate automation.
|
||||
|
||||
Start with the safe, read-only connection check: https://premiere-pro-mcp.com/blog/
|
||||
|
||||
### X / Bluesky
|
||||
|
||||
New practical guides for AI-assisted Adobe Premiere Pro workflows.
|
||||
|
||||
Compare the current native Assistant beta with a structured MCP path, then use focused backup, review, and delivery-QC checklists before relying on a workflow.
|
||||
|
||||
Free, MIT licensed, local-first: https://premiere-pro-mcp.com/blog/
|
||||
|
||||
### Community post
|
||||
|
||||
This open-source MCP server provides structured, local-first Adobe Premiere Pro workflows. The goal is not autonomous editing: it is making repetitive work inspectable, bounded, and easier to verify.
|
||||
|
||||
The guide hub covers the connection model, workflow boundaries, the current Adobe AI Assistant beta, project backups, visual review, and delivery QC. Feedback on capability boundaries, install friction, and real editor workflows is especially welcome once the verified guide URLs are live: https://premiere-pro-mcp.com/blog/
|
||||
|
||||
## Outreach email
|
||||
|
||||
**Subject:** A practical, local-first guide to AI workflows in Premiere Pro
|
||||
|
||||
Hi [name],
|
||||
|
||||
I published a short guide series for editors and post-production teams exploring AI-assisted Adobe Premiere Pro workflows. It focuses on a gap that gets missed in “AI video editing” coverage: how to inspect a project, constrain an automation, and verify a result instead of treating a prompt as proof.
|
||||
|
||||
MCP for Adobe Premiere Pro is a free, MIT-licensed local MCP server. The guides are useful even for readers who are comparing approaches, because they explain a clear safety and review model.
|
||||
|
||||
If it is relevant to your audience, the hub is here: https://premiere-pro-mcp.com/blog/
|
||||
|
||||
Thanks,
|
||||
MCP for Adobe Premiere Pro contributors
|
||||
|
||||
## 30-day distribution cadence
|
||||
|
||||
| Week | Primary action | Success signal |
|
||||
| --- | --- | --- |
|
||||
| 1 | Announce the guide hub on owned social channels and GitHub Discussions; link to the “what is” guide. | Referral visits and completed setup-guide clicks. |
|
||||
| 2 | Share a concrete inspect-plan-apply-verify example; link to the AI workflow guide. | Time on article and safe-check copy or docs clicks. |
|
||||
| 3 | Publish one short workflow recipe from a real, approved use case; link to the automation guide. | Qualified feedback and issue/discussion quality. |
|
||||
| 4 | Reach out individually to relevant editor, MCP, and post-production publications or newsletters. | Earned links and branded search impressions. |
|
||||
|
||||
## Measurement
|
||||
|
||||
- Tag each owned social link with a channel-specific UTM source and campaign such as `utm_source=linkedin&utm_medium=organic_social&utm_campaign=guides_launch`.
|
||||
- Treat a completed installation, `verify_premiere_connection`, and first successful supported tool call as stronger outcomes than impressions, likes, stars, or downloads.
|
||||
- Review Search Console query/impression data after enough crawl and ranking time; do not rewrite a guide solely because early rankings are sparse.
|
||||
|
||||
## Boundaries
|
||||
|
||||
- Do not characterize simulated UI/video assets as live Premiere proof.
|
||||
- Do not claim a tool works in every Premiere build; direct readers to the read-only connection check and capability inspection.
|
||||
- Do not imply that the local-first server changes the privacy terms of a chosen AI client.
|
||||
- This kit is for organic distribution. Paid campaigns are intentionally not activated: they need a defined budget, audience, destination, and conversion measurement plan.
|
||||
Reference in New Issue
Block a user