artifact-tracking

The Artifact Tracking skill provides a robust framework for managing software development workflows through token-efficient documentation. Designed for AI agents like Claude, this skill replaces verbose, pure-markdown tracking with a YAML-Markdown hybrid format. By leveraging machine-readable frontmatter, the system enables agents to perform surgical field-level updates and queries, which reduces token usage by 95-99% compared to traditional full-file parsing. It is an essential tool for engineering teams and autonomous agents aiming to maintain complex, multi-phase project structures without hitting context limits or incurring unnecessary costs.

• Enables CREATE, UPDATE, QUERY, VALIDATE, and ORCHESTRATE operations through a standardized command interface.

• Features structured metadata for automated task status updates, dependency management, and progress reporting.

• Supports phase-based project documentation including progress trackers, technical context notes, and PRD-aligned implementation plans.

• Facilitates session handoffs by allowing agents to ingest only the essential, machine-readable YAML state instead of entire project logs.

• Integrates with dedicated agents (tracker, query, validator) to ensure schema compliance and high-quality outputs across development lifecycles.

• Ideal for software engineers, project leads, and AI orchestrators handling long-running features or complex refactoring efforts.

• Input requirements: Follows specific file location patterns such as .claude/progress/[prd-name]/phase-N-progress.md.

• Best practices: Maintain one progress file per phase and use the embedded Task() function syntax for consistent state changes.

• Constraints: Strictly adheres to documentation policy specifications; schemas must be validated using the provided tools to ensure orchestration readiness.

• Always invoke the appropriate agent (e.g., artifact-tracker for updates, artifact-query for retrieval) to maximize efficiency and maintain artifact integrity.