agentleFS
full-aigc-pluginsfull-aigc-plugins/content-factory-plugin/skills/content-harness/SKILL.md

content-harness

Skill0 starsChanged 19 days agoApache-2.0View on GitHub

Content Factory's single plugin-local orchestration skill. Use for content production, editing, formatting, detection, delivery, or local Zhuque API Key setup. Resolve the requested task mode first, preserve supplied context, route only the stages needed for that mode, and leave canonical state/approval/delivery authority to the Runtime Kernel.

The file

Files
SKILL.md

Content Harness

Content Harness is the only plugin-local Skill. Vendor synchronization MUST NOT overwrite this directory.

Responsibilities

  1. Identify the requested run mode: full, edit, format, detect, repurpose, or deliver.
  2. Preserve topic, sources, channel, locale, audience, goal, content revision and account references already supplied.
  3. Resolve HostContext, SourceContext and ChannelIntent independently, then select a versioned ChannelProfile and ContentRecipe before channel-specific authoring. See platform routing.
  4. Produce a ContentBrief plus a bounded stage DAG using the runtime dispatcher.
  5. Report genuinely missing inputs. Never invent facts, accounts, sources, preferences, approval, detection results or delivery state.
  6. Hand candidate outputs to Runtime Kernel validation. A Skill cannot promote its own output to canonical state.

Mode routes

ModeRequired stages
fullbrief → research → outline → write → fact-check → edit → visual-plan → format → review
editedit → fact-check → review
formatformat → review
detectdetect → review
repurposerepurpose → fact-check → format → review
deliverdelivery-preflight → approval → deliver → delivery-verify

A simple request MUST NOT be inflated into the full pipeline. In particular, format does not authorize research, rewriting, paid detection or remote delivery.

Missing-input policy

  • full: require at least a topic or source reference.
  • edit, format, detect, repurpose: require a content revision/reference.
  • deliver: require content, channel and destination account.
  • Missing inputs are returned in missingInputs; assumptions remains empty unless the user explicitly approved an assumption.

Zhuque credential setup

When the user asks to configure the Zhuque API Key, open the bundled local setup page through content_factory_zhuque_setup if the MCP tool is available. In a source checkout, npm run zhuque:setup is the CLI fallback. Ask the user to enter the key in that page; never request, repeat, or place the key in chat, article files, or command arguments. content_factory_zhuque_status or npm run zhuque:status reports configuration presence only. A saved key does not prove API authentication or content detection; keep the normal consent, exact-revision, and real-response gates for detect mode.

Authority boundary

The harness may choose methods and request stages, but the Runtime Kernel owns:

  • immutable revisions and conflicts;
  • claims and fact guards;
  • budgets and credentials;
  • AI content detection platform validity;
  • trusted approval;
  • delivery intent, recovery and readback verification.

Generated visual production is delegated to Image Factory. Channel/profile-aware strategy selection is added by CF-043–CF-050 without creating additional plugin-local Skills.

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.