SuperAPI AI Directory

chanjing-content-creation-skill

Agents & models

View repository

Updated 10-15 17:33

About this skill

chanjing content creation skill

Cicada Mirror Content Creation Skill Package (top-level routing)

This file is only responsible for: trigger boundaries, routing disambiguation, unified execution constraints, and runtime contract summary.
Specific execution steps are subject to sub-skill *-SKILL.md as the source of truth; long-form details are subject to references/ as the source of truth.

Use When

Use this Skill when the user's needs involve any of the following:

  • Short video finished product (including one-click video composition, cartoon video, narration mashup)
  • Cicada Mirror voice capabilities (TTS, voice cloning)
  • Digital human capabilities (public/custom digital humans, lip sync, text-to-digital-human)
  • Cicada Mirror AI creation (text-to-image, text-to-video)
  • Cicada Mirror credential management (APP_ID/SECRET_KEY configuration, Token refresh, login guidance)

Do NOT Use

  • General needs unrelated to the Cicada Mirror platform (pure copywriting, non-Cicada Mirror editing tools, etc.)
  • The user is only discussing a plan and explicitly will not execute any Cicada Mirror API/script calls

Route Map (unique mapping)

User intent Routing target
One-click video composition, complete short video, make an xx video orchestration/chanjing-one-click-video-creation/chanjing-one-click-video-creation-SKILL.md
Cartoon short drama, anime multi-shot narrative orchestration/cartoon-video-creation/cartoon-video-creation-SKILL.md
Text-to-speech, speech synthesis products/chanjing-tts/chanjing-tts-SKILL.md
Voice cloning, timbre replication products/chanjing-tts-voice-clone/chanjing-tts-voice-clone-SKILL.md
Lip sync, lip-shape driven (non-cartoon) products/chanjing-avatar/chanjing-avatar-SKILL.md
Digital human narration, public digital human, pure narration video, cartoon digital human narration synthesis products/chanjing-video-compose/chanjing-video-compose-SKILL.md
Text-to-image, creative video products/chanjing-ai-creation/chanjing-ai-creation-SKILL.md
Custom digital human (upload material training) products/chanjing-customised-person/chanjing-customised-person-SKILL.md
Persona image, portrait LoRA, text-to-digital-human avatar products/chanjing-text-to-digital-person/chanjing-text-to-digital-person-SKILL.md
Credentials, APP_ID/SECRET_KEY, Token products/chanjing-credentials-guard/chanjing-credentials-guard-SKILL.md

Conflict Resolution (mandatory disambiguation)

When overlapping intents occur, determine by the following priority:

  1. Complete finished product (copy + storyboard + visuals + delivery) takes priority for routing chanjing-one-click-video-creation
  2. Cartoon/anime multi-shot narrative finished product takes priority for routing cartoon-video-creation
  3. Cartoon digital human narration synthesis (model=2) or digital human narration only, route chanjing-video-compose
  4. Existing video lip modification (non-cartoon) route chanjing-avatar
  5. When routing is uncertain, follow orchestration/orchestration-contract.md 's need_param branch minimal parameter supplementation

Root Workflow (minimal entry process)

Step Action Output Next step Failure branch
1 Credential availability check (when missing, go to chanjing-credentials-guard) ok or auth_required Step 2 auth_required: complete authentication first, then return to Step 2
2 Select a single target sub-skill according to the Route Map Target *-SKILL.md path Step 3 need_param: only supplement information required for routing
3 Open the target sub-skill and execute its Standard Workflow Sub-skill result Step 4 Execute according to the sub-skill failure branch
4 Return the artifact URL/path or standard error semantics ok/need_param/auth_required/upstream_error/timeout End When cross-product degradation is needed, obtain user confirmation first

Global Execution Rules

  • Only existing entry points of products/*/scripts/ and orchestration/*/scripts/ may be invoked.
  • Writing temporary scripts to replace existing CLIs is prohibited.
  • By default, do not download proactively; execute the download script only when the user explicitly requests "download/save locally".
  • Repeated tasks may reuse confirmed parameters, but when core inputs conflict, a new task must be started.
  • External side effects (network, disk writes, subprocesses, browser) must be consistent with the runtime contract.

Three-Layer Architecture

Tier Path Purpose
Top-level entry SKILL.md Routing, disambiguation, contract summary
L1 Common layer common/ Common authentication, HTTP wrapper, upload/download, logging
L2 Product layer products/ Single-product capabilities and standard scripts
L3 Orchestration layer orchestration/ Cross-product scenario orchestration

The dependency direction is fixed as:common <- products <- orchestration(reverse dependencies are prohibited).

Runtime Contract (Summary)

For the complete contract, see references/top-level-runtime-contract.md。

  • Credential source: project .env(CHANJING_APP_ID、CHANJING_SECRET_KEY)
  • Credential file:skills/chanjing-content-creation-skill/.env
  • Token cache:.env in CHANJING_ACCESS_TOKEN、CHANJING_TOKEN_EXPIRE_IN(refreshed by expiration time)
  • API base URL:${CHANJING_API_BASE:-https://open-api.chanjing.cc}
  • Key environment variables:CHANJING_APP_ID、CHANJING_SECRET_KEY、CHANJING_ACCESS_TOKEN、CHANJING_TOKEN_EXPIRE_IN、CHANJING_API_BASE、CHAN_SKILLS_DIR
  • External binaries:python3(all script entry points),ffmpeg/ffprobe(mainly used for the L3 final-video pipeline)
  • Side effect categories: outbound HTTPS, local writes, subprocesses, browser-guided login

Constraints

  • The script's disk-write boundary is constrained by the runtime contract (credential paths,--output-dir、output/、work/)。