agentleFS
imMamdouhaboammarimMamdouhaboammar/linkedin-animated-infographics/AGENTS.md

linkedin-animated-infographics

AGENTS.md14 starsChanged 3 months agoMITView on GitHub

This repository is an executable multi-host plugin ecosystem for evidence-safe LinkedIn infographics. Claude Code, Codex, and ChatGPT use package adapters around the same canonical product core. Do not treat skills/, agents/, research/, helper/, or tools/ as independent islands. Read helper/GUIDE.md before changing or running production behavior. Machine-readable authority: Run the strict runtime and host-package validators after structural changes: new-post is the parent workflow. Workers return artifacts to it; workers do not orchestrate peers directly. The complete path is: design-study ->…

The file

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

9 rules. About 0 are common, 3 are distinctive, and 6 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈3

    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
  1. Very specific:

    Read helper/GUIDE.md before changing or running production behavior.

    line 7
  2. Distinctive:

    Run the strict runtime and host-package validators after structural changes:

    line 24
  3. Very specific:

    After verified delivery, share-demo may run only with explicit user consent.

    line 45
  4. Very specific:

    Do not create copied Codex-specific product Skills.

    line 66
  5. Very specific:

    Do not package ignored research/upstreams/ clones.

    line 76
  6. Distinctive:

    Add or update a failing regression/contract test

    line 80
  7. Distinctive:

    Run the full suite and validators

    line 83
  8. Very specific:

    Update docs/manifest/version when public behavior changes

    line 84
  9. Very specific:

    Use masterone as the first user-facing entry for LinkedIn infographic work.

    line 109

≈ “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

Repository Agent Guide

This repository is an executable multi-host plugin ecosystem for evidence-safe LinkedIn infographics. Claude Code, Codex, and ChatGPT use package adapters around the same canonical product core. Do not treat skills/, agents/, research/, helper/, or tools/ as independent islands.

Start here

Read helper/GUIDE.md before changing or running production behavior.

Machine-readable authority:

  • helper/router.json: request to workflow/skills/agents/capabilities
  • helper/capabilities.json: capability owners and plugin-local defaults
  • helper/quality-gates.json: local creative/product gates
  • helper/artifacts.json: worker handoff artifacts
  • helper/modules.json: active public module manifest
  • research/capability-notes/gates.json: adopted research-derived runtime gates
  • architecture/plugin-graph.json: executable worker order and skill preloads
  • scripts/info_stories.py::load_catalog(): merged Info-stories registry
  • skills/info-stories/references/asset-source-policy.md: Lobe-first verified identity policy
  • skills/info-stories/references/typography-direction.md: intentional typography policy
  • compatibility/codex.json: OpenAI package/parity declaration
  • compatibility/antigravity.json: Antigravity package/parity declaration

Run the strict runtime and host-package validators after structural changes:

python3 scripts/ecosystem_doctor.py check
python3 scripts/validate_marketplace.py
python3 scripts/validate_codex_plugin.py
python3 scripts/validate_antigravity_plugin.py

Complete production path

new-post is the parent workflow. Workers return artifacts to it; workers do not orchestrate peers directly.

The complete path is:

design-study -> evidence-checker -> asset-curator -> creative-director -> story-architect -> palette-curator -> type-curator -> copy-compressor -> layout-composer -> caption-writer -> artboard-builder -> motion-director -> optional mascot-animator -> motion-engineer -> render-qa -> post-critic -> story-verifier

asset-curator owns verified identity sourcing. creative-director owns concept structure. type-curator owns intentional typography before copy fitting. Downstream workers preserve build/asset-plan.json and build/type-spec.json rather than selecting replacements.

Attention-bearing design copy follows hooked-design-copy, and a complete concept should produce a useful creative-payoff with clean-creative-structure rather than decorative spectacle.

After verified delivery, share-demo may run only with explicit user consent. Community publishing ends at an open contributor PR and requires maintainer manual review and merge.

Product defaults

  • Palette default: creative-attractive-restrained
  • Composition default: center-first
  • Identity default: Lobe-first after an exact user-supplied official asset. Supported named AI/tool identities resolve through Lobe; missing verified identity assets HOLD instead of being approximated
  • Typography default: intentional typography. User-specified or supplied render-safe fonts win; otherwise use a curated deterministic system direction. Remote font loading during capture is forbidden
  • Creative structure: one dominant visual anchor, explicit relationship, deliberate containment, and deliberate negative space. Repeated cards are used when repetition is the story
  • Alignment exceptions are valid for tables, UI mockups, code/terminal surfaces, timelines, Arabic/RTL flow, or documented reference DNA when they improve comprehension/fidelity
  • UI mockups that look real must be evidence-backed; conceptual UI must be identifiable when readers could mistake it for product proof

Host packaging

  • Claude Code: .claude-plugin/plugin.json + .claude-plugin/marketplace.json
  • Codex/ChatGPT: .codex-plugin/plugin.json + .agents/plugins/marketplace.json
  • Google Antigravity: plugin.json + .agents/plugins.json + .agents/skills.json + .agents/hooks.json + .agents/rules/*.md + .agents/agents/catalog.json + compatibility/antigravity.json
  • canonical bundled workflows: skills/
  • Codex repo-development helpers: .codex/config.toml + .codex/agents/*.toml
  • Antigravity agent definition generator: scripts/antigravity_agents.py

Do not create copied Codex-specific product Skills. The installed OpenAI plugin must not depend on the maintainer's .codex/config.toml or private MCP configuration.

Research gates

Research under research/ is active production guidance. Adopted runtime gates live in research/capability-notes/gates.json and are validated by:

python3 scripts/research_gates.py check

Do not package ignored research/upstreams/ clones. Keep source URL, inspected SHA, license, Adopt/Adapt/Reject decisions, local implementation references, shipping owners, and tests.

Change method

  1. Add or update a failing regression/contract test
  2. Implement the minimum coherent change
  3. Run focused tests
  4. Run the full suite and validators
  5. Update docs/manifest/version when public behavior changes
  6. Merge only after exact-head host validation, available marketplace smoke checks, repository checks, and review findings are clean

Minimum deterministic gate:

python3 -m unittest discover -s tests -v
python3 -m compileall -q scripts tools skills/svg-mascot-animator/scripts
python3 scripts/info_stories.py check
python3 scripts/ecosystem_router.py check
python3 scripts/research_gates.py check
python3 scripts/plugin_graph.py check
python3 scripts/ecosystem_doctor.py check
python3 scripts/demo_gallery.py check
python3 scripts/validate_marketplace.py
python3 scripts/validate_codex_plugin.py
python3 scripts/validate_antigravity_plugin.py

Safety

Never commit credentials, private MCP data, or user secrets. Do not fabricate claims, metrics, logos, UI states, testimonials, product behavior, or identity assets to satisfy a visual slot. Report browser/render environment blockers separately from non-browser validation. Do not claim OpenAI public-directory publication until the external submission, review, and publication steps have actually completed.

MasterOne front door

Use masterone as the first user-facing entry for LinkedIn infographic work. It owns first-run onboarding, persistent project preferences, and route selection. The persistent profile lives at .linkedin-infographics/profile.json and may include copyright/footer text, typography, references, logos, mascots, language, audience, and reusable production preferences.

Do not make the masterone worker coordinate peer workers. The parent/client transfers a ready request to the route in helper/router.json; new-post remains the canonical complete-production parent workflow. Ask only for materially missing inputs and never invent copyright, attribution, fonts, official identity assets, mascot identities, or reference intent.

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.