DOCSEARCHbyAlgolia
Submit Docs

Awesome Codex Skills

repository·master·Indexed 12 days ago

https://github.com/composio-community/awesome-codex-skills

A curated repository of practical skills for automating workflows across the Codex CLI and API. It includes modular instructions and evaluation frameworks for skills such as Knowledge Capture, Meeting Intelligence, Research & Documentation, and Spec to Implementation, with specific testing guidelines for Codex models Haiku, Sonnet, and Opus.

Tokens
557.2K
Snippets
1.9K
Records
3.3K
Agent score
96%

What's inside Awesome Codex Skills

  1. Alternative accounting and invoicing integrations

    master

    Since native Wave Accounting tools are currently unavailable in Composio, you can use the following supported alternatives for accounting and invoicing automation:

    • Stripe: For payment processing, invoicing, and subscription management.
    • Zoho Invoice: For invoice creation, payment tracking, and contact management.
    • QuickBooks: For a full accounting suite including invoicing and expense tracking.
    • FreshBooks: For cloud accounting with time tracking and invoicing.

  2. Use the brand-guidelines skill

    master

    The brand-guidelines skill applies OpenAI's brand identity, including specific colors and typography, to artifacts to ensure they match the Codex/OpenAI look-and-feel. Use this skill when visual formatting, company design standards, or specific brand colors are required for an output.

    name: brand-guidelines
description: Applies OpenAI's brand colors and typography to any artifact that should match the Codex/OpenAI look-and-feel. Use it when brand colors or style guidelines, visual formatting, or company design standards apply.
license: Complete terms in LICENSE.txt

  4. Categorize and prioritize requirements

    master

    When parsing requirements, group them into the following categories:

    • Functional: System behaviors, user capabilities, and data operations.
    • Non-Functional: Performance targets, security, scalability, availability, and compliance.
    • Constraints: Technical, business, or timeline limitations.

    Identify priority levels to map implementation phases:

    • P0 (Critical): "Must have", "Critical"
    • P1 (Important): "Should have", "Important"
    • P2 (Nice to have): "Could have"
    • P3 (Future): "Won't have"

  6. What are Codex Skills and how do they work?

    master

    Codex Skills are modular, self-contained packages that extend the capabilities of the Codex agent by providing specialized knowledge, workflows, and tools. They act as "onboarding guides" for specific domains, transforming a general-purpose agent into a specialized one.

    Core Functions

    1. Specialized workflows: Multi-step procedures for specific domains.
    2. Tool integrations: Instructions for working with specific file formats or APIs.
    3. Domain expertise: Company-specific knowledge, schemas, or business logic.
    4. Bundled resources: Scripts, references, and assets for complex tasks.

    The Progressive Disclosure Model

    To manage the context window efficiently, skills use a three-level loading system:

    1. Metadata (name + description): Always in context (~100 words).
    2. SKILL.md body: Loaded only when the skill triggers (<5k words).
    3. Bundled resources: Loaded by Codex as needed (unlimited capacity via scripts or targeted reference reading).

  8. Understand Knowledge Capture evaluation scenarios

    master

    Knowledge Capture evaluations use specific JSON files to test how the skill handles different documentation tasks:

    conversation-to-wiki.json

    Goal: Tests capturing conversation content as a how-to guide for a team wiki. Key Behaviors to Verify:

    • Extraction of steps, gotchas, and best practices.
    • Identification of content as a How-To Guide.
    • Proper structuring with sections: Overview, Prerequisites, Steps, and Troubleshooting.
    • Successful search for the team wiki location.
    • Preservation of technical details like commands and configs.

    decision-record.json

    Goal: Tests capturing architectural or technical decisions with full context. Key Behaviors to Verify:

    • Extraction of decision context, alternatives, and rationale.
    • Adherence to the decision record structure: Context, Decision, Alternatives, and Consequences.
    • Capture of both selected and rejected options with reasoning.
    • Correct placement in a decision log or ADR database.
    • Linking to related technical documentation.

  9. Best practices and pitfalls for Landbot automation

    master

    To ensure reliable Landbot automation, follow these guidelines:

    • Search First: Never hardcode tool slugs or arguments. Always call RUBE_SEARCH_TOOLS first to get current schemas.
    • Verify Connection: Always confirm RUBE_MANAGE_CONNECTIONS returns an ACTIVE status.
    • Schema Compliance: Use exact field names and types as defined in the search results.
    • Memory Parameter: Always include memory: {} in RUBE_MULTI_EXECUTE_TOOL calls.
    • Session Management: Reuse the same session_id within a single workflow, but generate a new one for entirely new workflows.
    • Pagination: When fetching results, check for pagination tokens and continue fetching until all data is retrieved.

  10. Best practices and pitfalls for Clickmeeting automation

    master

    When automating Clickmeeting via Rube MCP, follow these guidelines to avoid common errors:

    • Dynamic Discovery: Always call RUBE_SEARCH_TOOLS first. Tool schemas change, so hardcoding slugs or arguments will lead to failures.
    • Connection Verification: Always verify that RUBE_MANAGE_CONNECTIONS returns an ACTIVE status before execution.
    • Schema Strictness: Use exact field names and types as defined in the search results.
    • Mandatory Memory: The memory parameter in RUBE_MULTI_EXECUTE_TOOL is mandatory; pass {} if no memory is needed.
    • Session Management: Reuse session IDs within a single workflow, but generate new ones for entirely new workflows.
    • Pagination: When fetching results, check for pagination tokens and continue fetching until the full dataset is retrieved.

  11. Visual Styling Features

    master

    The brand-guidelines skill includes several automated styling behaviors:

    • Smart Font Application: Automatically detects text type (heading vs. body vs. code) and applies the corresponding font and weight.
    • Smart Color Selection: Automatically selects appropriate text colors (Charcoal or Mist) based on the background color to maintain contrast.
    • Shape and Accent Cycling: Non-text shapes use accent colors, cycling through the sequence: OpenAI Green $\rightarrow$ Azure $\rightarrow$ Graphite to maintain visual interest while staying on-brand.

  12. Common workflows and patterns for Google Slides automation

    master

    Recommended Patterns

    • Markdown-first: Use GOOGLESLIDES_CREATE_SLIDES_MARKDOWN for rapid generation from structured text.
    • Template-based: Use GOOGLESLIDES_PRESENTATIONS_COPY_FROM_TEMPLATE to duplicate a styled template, then use GOOGLESLIDES_PRESENTATIONS_BATCH_UPDATE to populate it with content.
    • Inspect then Modify: Use GOOGLESLIDES_PRESENTATIONS_GET to find specific pageObjectIds or element structures, then use GOOGLESLIDES_PRESENTATIONS_BATCH_UPDATE for targeted edits.
    • Thumbnail Export: List page IDs with GOOGLESLIDES_PRESENTATIONS_PAGES_GET, then use GOOGLESLIDES_PRESENTATIONS_PAGES_GET_THUMBNAIL for previews.
    • Sharing: Combine with the googledrive toolkit's GOOGLEDRIVE_ADD_FILE_SHARING_PREFERENCE to share presentations after creation.