bcal-workflow: Agent-agnostic Workflow Skill Suite for Claude Code

An agent-agnostic workflow skill set for Claude Code and Codex, offering session lifecycle management, diary logging, review notifications, and other features. It supports installation via the plugin marketplace or manual integration.

bcal-workflow is part of the bcal-agent-plugins marketplace launched by Blue Collar AI Labs. It focuses on solving pain points of AI programming assistants in real engineering workflows: messy session management, context loss, unsynchronized team collaboration information, etc. Unlike traditional development tools targeting specific frameworks or languages, bcal-workflow adopts an "agent-agnostic" design philosophy—it is not tied to a specific AI model or IDE. Instead, it is based on universal command interfaces and file conventions, and can run in Claude Code, OpenAI Codex, or other AI assistant environments that support similar interaction modes.

The project provides five main workflow skills:

Command: /bcal-workflow:start-session

Function: Intelligently resume or create a new development session. The system checks the NEXT_SESSION.md file—if it exists, it restores the context and to-do items from the last session; if not, it creates a brand-new session record. This solves the problem of context breakage in multi-round development, ensuring the AI assistant can "remember" previous progress.

Command: /bcal-workflow:end-session-gracefully

Function: A standardized session wrap-up process that automatically performs the following operations:

• Commit code changes in the current workspace to Git
• Push to the remote repository
• Update the project status file
• Generate a handoff prompt for easy resumption in the next session or for team members to take over

This design embodies professional software engineering practices: each work unit should have a clear start and end, and state changes should be recorded and synchronized.

Command: /bcal-workflow:write-to-diary

Function: Append the summary of the current session to the DIARY.csv file. Unlike Git commits that record code changes, the diary logs "meta-information" such as decision-making processes, obstacles encountered, and solution ideas—these are crucial for long-term project maintenance and team knowledge transfer. The CSV format is chosen for easy subsequent data analysis and visualization, such as calculating the development time of each module or identifying high-frequency problem areas.

Command: /bcal-workflow:notify-proof

Function: Automatically send a Slack notification when a document is submitted to Proof (a review system). This requires configuring the SLACK_PROOF_WEBHOOK_URL environment variable to point to a Slack Incoming Webhook.

This feature bridges the gap between AI-assisted development and team collaboration tools—AI assistants are no longer just code generators, but notification nodes that can actively trigger team workflows.

Command: /bcal-workflow:live-transcript

Function: Capture group discussion content from Notion's live transcription page and generate a comprehensive answer or summary. This requires pre-configuring the Notion MCP (Model Context Protocol) server and API integration token.

Application scenarios include: extracting key points after remote meetings, aggregating ideas from brainstorming sessions, and recording conclusions from technical discussions.