Skill: codex-claudeception-migration

When to use

• The repository already contains .claude/skills or Claude-specific instructions and must be migrated to Codex.
• The team wants to preserve post-task review -> skill extraction, without depending on Claude hooks.
• You need an MVP first (directories + docs + manual flow) before adding automation.

Inputs

• Existing Claudeception assets (README, SKILL files, templates, scripts)
• Target Codex constraints (writable paths, AGENTS rules, available tools)
• Migration scope (MVP only or full migration)

Steps

1. Identify non-portable parts: Claude hooks, Claude-specific tool names, Claude command entrypoints.
2. Map structure: .claude/skills/* -> skills/*; templates -> docs/ or scripts/.
3. Add mandatory post-task review and quality gates in AGENTS.md.
4. Standardize skill format with frontmatter + when/steps/validation/anti-patterns.
5. Create one real migrated skill and validate usability before investing in automation.

Validation

• Checkpoint 1: project includes AGENTS.md, skills/skill-template.md, and migration checklist docs.
• Checkpoint 2: at least one formal skill exists at skills/<topic>/SKILL.md.
• Checkpoint 3: that skill has explicit triggers, executable steps, and observable verification points.
• Checkpoint 4: later tasks can retrieve and reuse the skill by scenario match.

Anti-patterns

• Copying Claude hook config without Codex-executable replacement.
• Overly generic descriptions that cannot be matched by keyword/semantics.
• Extracting one-off experience as a skill with no reuse value.
• Implementing complex automation before validating examples.

Example

• Scenario: migrate Claudeception into Codex by building codexception first.
• Output: directory scaffold, AGENTS review rules, migration checklist, and one formal migration skill.
• Outcome: team can start manual consolidation now and automate later.