Agent skill
skill-creator
Create or update Codex skills (SKILL.md plus optional scripts/references/assets). Use when asked to design a new skill, update an existing skill, or validate/package skills.
Install this agent skill to your Project
npx add-skill https://github.com/therealityreport/THB-BBL/tree/main/skills/skill-creator
SKILL.md
Skill Creator
Skills are modular packages that extend AI agent capabilities with specialized knowledge, workflows, and tools. They follow the open Agent Skills specification and work across Codex and other compatible agents.
Core Principles
Concise is Key
The context window is shared. Only add information Codex does not already have. Prefer concise examples over verbose explanations.
Degrees of Freedom
Match specificity to task fragility:
- High freedom (text instructions): Multiple approaches valid, context-dependent
- Medium freedom (pseudocode/parameterized scripts): Preferred pattern exists, some variation acceptable
- Low freedom (specific scripts): Operations fragile, consistency critical
Skill Anatomy
skill-name/
├── SKILL.md (required)
│ ├── YAML frontmatter (name, description)
│ └── Markdown instructions
└── Bundled Resources (optional)
├── scripts/ - Executable code
├── references/ - Documentation loaded as needed
└── assets/ - Files used in output (templates, images)
SKILL.md Structure
Frontmatter (YAML):
name: Skill identifier (kebab-case, max 64 chars, must match directory name)description: What it does AND when to trigger it (max 1024 chars). This is the primary trigger mechanism; include all "when to use" info here, not in the body.metadata(optional): Key-value pairs likeauthor,versioncompatibility(optional): Environment requirements (max 500 chars)license(optional): License name or reference to LICENSE file
Body (Markdown): Instructions loaded after skill triggers. Keep under 500 lines.
Bundled Resources
scripts/: Executable code for deterministic, repeatable tasks. Token-efficient; may execute without loading into context.
references/: Documentation Codex reads as needed. Keeps SKILL.md lean. For files >10k words, include grep patterns in SKILL.md.
assets/: Files used in output (templates, logos, fonts). Not loaded into context, copied or modified in final output.
Do NOT Include
- README.md, CHANGELOG.md, INSTALLATION_GUIDE.md
- User-facing documentation
- Setup/testing procedures
Progressive Disclosure
Three-level loading system:
- Metadata (~100 words) - Always in context
- SKILL.md body (<5k words) - When triggered
- Bundled resources (unlimited) - As needed
When approaching 500 lines, split into reference files. Reference all split files from SKILL.md with clear guidance on when to read them.
Pattern examples:
## Advanced features
- **Form filling**: See [FORMS.md](references/FORMS.md)
- **API reference**: See [REFERENCE.md](references/REFERENCE.md)
For multi-domain skills, organize by domain so Codex only loads relevant context.
Creation Process
Step 1: Understand with Examples
Gather concrete usage examples. Ask:
- What functionality should this skill support?
- What would a user say to trigger this skill?
- Can you give example requests?
Step 2: Plan Reusable Contents
For each example, identify:
- What scripts would help (repeated code)?
- What references would help (schemas, docs)?
- What assets would help (templates, boilerplate)?
Step 3: Initialize
Run the initialization script:
python3 skills/skill-creator/scripts/init_skill.py <skill-name> --path <output-directory>
Creates skill directory with SKILL.md template and example resource directories.
Step 4: Edit
Frontmatter:
- Write clear, comprehensive
descriptionwith all trigger scenarios - Example: "Comprehensive document creation and editing. Use when working with .docx files for: (1) Creating documents, (2) Editing content, (3) Tracked changes, (4) Adding comments"
Body:
- Write for another Codex instance to use
- Include non-obvious procedural knowledge
- Use imperative/infinitive form
- See
references/workflows.mdfor workflow patterns - See
references/output-patterns.mdfor output templates
Resources:
- Implement scripts, references, assets identified in Step 2
- Test scripts by running them
- Delete unused example files
Step 5: Package
python3 skills/skill-creator/scripts/package_skill.py <path/to/skill-folder>
Validates and creates .skill file for distribution.
Step 6: Iterate
Use the skill on real tasks, identify struggles, update accordingly.
Recommended Agent Skills
Expand your agent's capabilities with these related and highly-rated skills.
bbl-senior-architect
Use for THB-BBL architecture decisions, ADR-ready tradeoff analysis, dependency boundaries, and long-term system impact across request parsing, team building, exports, scheduler, and agent tool boundaries. Do not use for routine implementation where design is already fixed.
bbl-senior-qa
Use for THB-BBL risk-tiered validation planning, regression prevention, and release verification for changed behavior across request analysis, team assignment, exports, and scheduler flows. Do not use for implementation-only tasks.
bbl-senior-devops
Use for THB-BBL CI/CD reliability, deployment hardening, rollback-readiness, environment safety, and operational gates. Do not use for feature-only coding without operational impact.
bbl-senior-backend
Use for THB-BBL backend/API/schema/security changes across Next.js route handlers, TypeScript services, Prisma schema, and legacy Python backend modules when downstream compatibility must be controlled. Do not use for UI-only work.
bbl-senior-frontend
Use for THB-BBL UI-first Next.js App Router implementation, accessibility, rendering/state behavior, and frontend performance when backend contracts are stable. Do not use when backend/API/schema contracts need to move.
repo-structure-reviewer
Audit a repository's structure and propose a safe, approval-gated reorganization plan. Use when asked to review repo anatomy, propose folder changes, or apply an approved reorg with rollback.
Didn't find tool you were looking for?