work-logs-knowledge-log

The work-logs-knowledge-log skill is designed to act as a documentation capture tool during AI-assisted development sessions. As an AI agent works on a repository, it inevitably identifies nuances, architectural intents, and undocumented business logic that are not immediately apparent from source code alone. This tool allows the agent to extract these findings and persist them as structured markdown knowledge logs for future reference, bridging the gap between raw code and human-readable project context. It is primarily intended for developers and technical leads who need to maintain institutional knowledge within the repository environment.

• Automatically generates timestamped knowledge log files in the ~/.claude/knowledge/${repo_name}/ directory.

• Prioritizes documenting the delta between inferred code logic and actual user business requirements or constraints.

• Captures implicit rules, operational practices, and specific design rationales that explain the 'why' behind certain code structures.

• Uses standard file system tools (Read, Write, Grep, Glob) to inspect code and persist documentation files.

• Supports standardized date-time prefixing based on the Asia/Tokyo timezone to ensure chronological consistency.

• Trigger this skill when encountering discrepancies between code functionality and project requirements or during discovery phases of a task.

• Use the recorded logs as a basis for identifying areas where official project documentation needs expansion or correction.

• Ensure the date-time format follows the +%Y-%m-%d_%H-%M convention for file organization.

• Rely on this skill to record technical debt, specific business logic exceptions, and architectural decisions that are not captured in commit messages or standard README files.

• The tool is constrained to operate within the defined ~/.claude/knowledge path, ensuring that knowledge accumulation remains localized to the agent's workspace.