subagent-creator
Design bounded sequential lenses of codenavi-agent with a single deliverable, explicit evidence packet, minimal tools and verifiable verdict. Use for create verifier, parity auditor, isolated review procedure or specialized work unit. Do NOT use for creating, register or spawn agent identities; use skill-architect for a reusable skill and cursor-subagent-creator for Cursor work-unit guidance.
- License 许可
- CC-BY-4.0
- 最近更新
- 2026-11-14
- Stars
- ★ 1
- Forks
- 0
01 前言 Front Matter
- name
- subagent-creator
- license
- CC-BY-4.0
- metadata
- { "author": "Tech Leads Club", "version": "1.1.0", "owner_agent": "codenavi-agent" }
- description
- Design bounded sequential lenses of codenavi-agent with a single deliverable, explicit evidence packet, minimal tools and verifiable verdict. Use for create verifier, parity auditor, isolated review procedure or specialized work unit. Do NOT use for creating, register or spawn agent identities; use skill-architect for a reusable skill and cursor-subagent-creator for Cursor work-unit guidance.
02 资源文件 10
- 参考文档.skill-meta.json120 B
- 参考文档PERMISSIONS.yaml388 B
- 参考文档POLICY.md189 B
- 参考文档references/lens-contract.md1.0 KB
- 参考文档references/lenses/retained/legacy-contract.md260 B
- 参考文档references/ORIGINAL_SKILL.md6.9 KB
- 参考文档references/package-notes.md254 B
- 参考文档references/provenance.md942 B
- 参考文档references/tlc/UPSTREAM.md6.9 KB
- 参考文档references/usage-and-verification.md917 B
03 SKILL.md 正文
subagent-creator
Governance: CodENavi Full Workspace v2. Skill of codenavi-agent; inherit AGENTS.md.
Lifecycle: BRIEFING → RECON → PLAN → EXECUTE → VERIFY → DEBRIEF.
Workflow
- Discover the work before choosing a role: one deliverable; judging or producing; which information must be excluded to make the judgment meaningful.
- Classify the work as audit, research or execution. Audit reads supplied evidence only; research may use approved read/search tools; execution may write or run tools only in authorized scope. Shell execution can write even without an edit tool.
- Read lens contract and choose the smallest tool set. A Markdown tool list is an instruction boundary, not proof of runtime enforcement.
- Write one lens under
00_skills/(category)/host-skill/lenses/name.md. State owner codenavi-agent, inputs and excluded information, numbered observable criteria, exact output and stop rule. - Use parity lens for migration result comparison. Keep diff, author, commit history and intent out of its evidence packet.
- Verify the file/link and perform the bounded lens pass in this thread. Report context contamination, missing evidence and remaining limitations; never claim true independent context from a sequential self-review.
Boundaries
Never create .devin/agents, AGENT.md profiles, agent: bindings or agents registries. Existing AGENTS.md stays authoritative and unchanged. Source runtime claims about background tools, default models and builtin names are host-specific, not guarantees here. For audit: missing evidence means INCONCLUSIVO, not approval. List what could not be checked; do not fix or suggest fixes in a parity verdict. For a repository scan, return only requested repo/path:line matches unless synthesis was requested. Source: IMG_2700–IMG_2704. Output: a linked lens procedure and truthful verdict, not a new assistant identity.
Examples and failure handling
Example: “create a parity verifier” → a local lens with antes/depois evidence and enumerated criteria; “scan this fleet for library X” → bounded repo/path:line matches. Failure: missing runtime enforcement or contaminated context → disclose the limitation, never add another identity.
Retained resources
- Read retained ORIGINAL_SKILL only for its matching mode; this entry and root governance take precedence.
- Read retained package-notes only for its matching mode; this entry and root governance take precedence.
- Read retained provenance only for its matching mode; this entry and root governance take precedence.
- Read retained tlc only for its matching mode; this entry and root governance take precedence.
- Read retained usage-and-verification only for its matching mode; this entry and root governance take precedence.
04 同仓库更多 Skills
workspace-brain
Operate the CodENavi workspace brain MCP and CLI — cloud DeepSeek RAG, disk corpus at WORKSPACE_BRAIN_ROOT, Jev URL curation, and event logging for Cursor/Codex/OpenCode/Claude. Use when ingesting repo knowledge, asking the brain, mapping web seeds, or recording AI-driven file changes. Do NOT use for local Ollama RAG or storing API keys in git.
workbench-profile-resolver
Provide enterprise-grade capability for `workbench-profile-resolver` inside AEOS v1.1. Use when the user requests workbench profile resolver or invokes workbench-profile-resolver. Do NOT use for unrelated tools or unapproved remote writes.
wordpress-expert
Staff-level WordPress engineering super-skill for one-shot site mapping, clean-code AI-assisted project creation/modification, the complete installed plugin universe, governed remote wp-admin operations, themes/blocks/layout, third-party integrations and production-safe changes. Use when the user requests wordpress expert or invokes wordpress-expert. Do NOT use for unrelated languages, domains or production deployment.
whatsapp-simulator
Generate synthetic WhatsApp-style conversation fixtures for testing conversational systems, support flows and automation. Use when the user requests whatsapp simulator or invokes whatsapp-simulator. Do NOT use for unrelated tools or unapproved remote writes.
web-quality-audit
Assimilated TLC/web-quality multi-area audit. Review performance, accessibility, SEO and best practices in one pass. Triggers: audit my site, review web quality, check page quality, optimize my website across areas. Do not use for a single-area job (core-web-vitals, ai-seo, best-practices, perf-lighthouse).
voiceai
Provide an evidence-backed voice interface for AEOS. Activating this skill is treated as explicit current-session consent for `voiceai` to record user audio or transcribe audio from user-authorized audio/video media within the declared scope. The skill can: Use when the user requests voiceai or invokes voiceai. Do NOT use for unrelated tools or unapproved remote writes.