agentleFS
celeryhqceleryhq/simplified-ai/AGENTS.md

simplified-ai

AGENTS.md4 starsChanged 4 months agoMITView on GitHub

Operational guidance for AI coding assistants (Codex, Claude Code, Cursor) using this plugin. Simplified for AI lets agents operate Simplified's image/video generation, assets, brand context, marketing projects, and social media, then compose those platform primitives into marketer workflows — through: - Skills (skills/) — SKILL.md workflow files that teach the agent sequencing, terminology, and safety behavior. - A hosted MCP connector — the tools the skills drive, served at https://apikit.simplified.com/mcp (OAuth-secured). Declared in .mcp.json.

The file

We split this AGENTS.md into the rules it gives an agent

4 rules. About 0 are common, 0 are distinctive, and 4 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈0

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 4

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    All three are the same Simplified account.

    line 97
  2. Very specific:

    Resolve scope before scoped work: Use api_getWorkspaceInfo as whoami when workspace/teamspace context is named or uncertain.

    line 104
  3. Very specific:

    Accountless drafts are supported: If connected accounts are empty, continue an authorized action: "draft" request without account_ids.

    line 121
  4. Very specific:

    When an AI agent commits to this repo, include a Co-Authored-By: line with the agent model's name.

    line 125

≈ “Common” is an estimate: at least 1 in 5 of the 40,246 AGENTS.md 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.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
AGENTS.md

AGENTS.md — Simplified for AI

Operational guidance for AI coding assistants (Codex, Claude Code, Cursor) using this plugin.

Project overview

Simplified for AI lets agents operate Simplified's image/video generation, assets, brand context, marketing projects, and social media, then compose those platform primitives into marketer workflows — through:

  • Skills (skills/) — SKILL.md workflow files that teach the agent sequencing, terminology, and safety behavior.
  • A hosted MCP connector — the tools the skills drive, served at https://apikit.simplified.com/mcp (OAuth-secured). Declared in .mcp.json.

Plugin structure

simplified-ai/
├── .mcp.json                 # hosted MCP connector (OAuth) — shared by all clients
├── .claude-plugin/plugin.json   # Claude Code manifest
├── .codex-plugin/plugin.json    # Codex / ChatGPT Apps manifest
├── skills/                   # SKILL.md workflows (Agent Skills spec)
│   ├── manage-assets/
│   ├── generate-image/
│   ├── edit-image/
│   ├── transcribe-media/
│   ├── generate-voiceover/
│   ├── repurpose-video/
│   ├── generate-video/
│   ├── simplified-workspace/
│   ├── simplified-social/
│   ├── manage-brand/
│   ├── manage-projects/
│   ├── simplified-project-management/
│   ├── simplified-cli/
│   ├── simplified-workflows/
│   ├── simplified-agents/
│   ├── social-content-planner/
│   ├── cross-platform-campaign/
│   ├── content-repurposer/
│   ├── evergreen-content-engine/
│   ├── local-business-marketing/
│   ├── creative-testing/
│   ├── social-performance-analyst/
│   └── campaign-review/
├── assets/                   # brand icon + logo
└── evals/                    # test cases + runnable I/O harness

Skills

SkillPurposeTools
manage-assetsSearch/list, import/upload, inspect, and reuse workspace mediaapi_listAssets, api_getAsset, intake tools
generate-imageText-to-image generation (Flux, Gemini/Imagen, GPT Image, Ideogram, …); saves as a reusable assetapi_generateImage
generate-videoModel-aware AI video generation and completion handlingapi_getModelFields, api_generateVideo, api_getVideoVariation
edit-imageBackground removal/replacement, restoration, upscaling, masked edits, and output verificationimage-processing media tools
transcribe-mediaEvidence-led transcription and SRT/VTT exportstranscription media tools
generate-voiceoverLocale-aware voice discovery and reusable narrationvoice media tools
repurpose-videoRender source-grounded highlights/custom excerpts, preserve timestamps, and prepare draftsrepurpose project and clip tools
simplified-workspaceAuthenticated identity, workspace settings, and safe teamspace discoveryworkspace and teamspace tools
simplified-socialDraft / schedule / queue posts, timed auto-comments, and analytics across 13 platformssocial_*
manage-brandEvidence-led brand kits and reusable brand contextbrand-kit and context-document tools
manage-projectsMarketing projects, deliverables, assignments, and exportsproject and item tools
simplified-project-managementBoards, tasks, subtasks, dependencies, assignees, tags, custom fields, commentspm_* (simplified-full connector)
simplified-cliThe smp command line across PM, media, assets, social, and smp serveCLI, not MCP
simplified-workflowsBuild, publish, run, and control multi-step workflowsSeparate automation connection, flows_*
simplified-agentsBuild, change, schedule, and publish Simplified AI agents, including website chat agentsSeparate automation connection, agents_*
social-content-plannerGoal-led weekly and monthly content calendarsaccounts, analytics, drafts, scheduling
cross-platform-campaignCoordinated channel-native campaign rolloutsimage generation + social
content-repurposerSource content into channel-native post sequencesdrafts + optional image generation
evergreen-content-engineDurable content territories, franchises, content bank, and renewal loopaccounts, analytics, drafts
local-business-marketingVerified local and Google Business programsaccounts, assets, drafts, platform settings
creative-testingControlled creative experiments and reusable learninganalytics, generation, drafts
social-performance-analystKPI, trend, post, and audience analysis with next actionssocial analytics
campaign-reviewDraft QA, revisions, and stakeholder review bundlesdrafts + review bundles

The outcome-driven skills compose the platform operators. Workspace identity establishes the scope for every other operator. Image and video generation with asset storage returns permanent asset IDs for simplified-social.media; brand and project skills provide reusable context and operational handoffs. Workflow skills orchestrate those primitives around marketer jobs without duplicating API mechanics.

MCP servers

Three connectors, configured in .mcp.json:

ServerURLServes
simplifiedhttps://apikit.simplified.com/mcpcontent, assets, brand, projects, social (api_, media_, notify_, social_)
simplified-automationhttps://apikit.simplified.com/automation/mcpagents and workflows (agents_, flows_, smp_callApi)
simplified-fullhttps://apikit.simplified.com/full/mcpeverything above plus project management (pm_*); used by simplified-project-management

All three are the same Simplified account. Each is OAuth-secured — the client walks the OAuth flow; no API key to set. The hosted ChatGPT app connects only to simplified, which is why the automation skills are left out of that bundle. On a 401, the client refreshes its token automatically (server emits the standard challenge).

Key conventions (agent behavior)

  • Confirm before spending credits. Image/video generation and social publishing consume credits / post to live accounts — confirm when intent is ambiguous.
  • Resolve scope before scoped work. Use api_getWorkspaceInfo as whoami when workspace/teamspace context is named or uncertain. Resolve names to exact numeric IDs, then pass space_id on every downstream tool call in that scoped task.
  • Draft before publish. For social posts, create a draft and show it before scheduling/queuing. Publish only within explicit user authorization; reuse approval for the same content, accounts, timing, and comments instead of asking again.
  • Keep agency reviews client-isolated. Resolve the client's workspace/teamspace, carry its space_id through every account, draft, update, and review-bundle call, and create a separate bundle per client and campaign. Never mix cross-client IDs.
  • "Post now" uses the available publishing action add_to_queue. Do not promise immediate delivery; report the actual timing returned by the service. There is no separate immediate publish action; the action enum is schedule | add_to_queue | draft.
  • Auto-comments are post-relative. comments[].delay is a nonnegative number of seconds after the post publishes, not after the preceding comment. Convert “first comment after X minutes” to X * 60, and include the comment text and delay in the publication plan the user authorizes.
  • Prefer asset IDs for Simplified media. Generated file URLs can expire. Social media accepts asset UUIDs, supported URLs, or URL/thumbnail objects. Posting account_ids is string[], while analytics account_id is integer; do not conflate these identifiers.
  • Present media using client capabilities. Use a native media preview when the client supports it and the user wants to review the result. Otherwise show a usable link. Refresh expired asset URLs; do not download remote media merely to bypass client display restrictions. Review bundles and other web pages remain links.
  • Accountless drafts are supported. If connected accounts are empty, continue an authorized action: "draft" request without account_ids. Scheduling, queueing, and account-specific analytics require the relevant connected accounts.

Commit attribution

When an AI agent commits to this repo, include a Co-Authored-By: line with the agent model's name.

Media contracts

Route asset discovery/intake/reuse to manage-assets. Use current generation schemas: image parameters.reference_images takes asset IDs or URLs; video normalized slots take asset UUIDs. Current middleware waits for generation completion; continue only returned non-terminal jobs. Installed skills do not change deployed tool availability. Do not use historical eval tool counts as current capabilities.

Download now (ZIP)

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
What did you use it for? e.g. Reviewing pull requests in a Rails app

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.