research-context-compressor

To save project context for future AI agents, this Skill generates a compact, machine-readable workspace manifest under the .research/ directory at your project root. It reads inputs like README files and design briefs to create project_manifest.yml, experiment_matrix.yml, and data_dictionary.yml.

A research manifest is a set of standardized YAML files that summarize a project's structure, experiments, and data dictionary. You need to create one when compressing project context to orient long-running workflows or when researchers switch contexts and AI agents take over tasks.

You create a .research workspace manifest by triggering the Skill to compress your project context. It reads available inputs such as README files, design briefs, and relevant directories, then writes project_manifest.yml, experiment_matrix.yml, and data_dictionary.yml to the .research/ folder.

Yes, this context compression approach works with any research repository structure that contains readable inputs like README files, design briefs, and relevant directories. It requires no specific dependencies and extracts project orientation data to build the manifest.

A .research workspace manifest includes project_manifest.yml, experiment_matrix.yml, and data_dictionary.yml as core files. It also generates run_log.md and open_questions files as applicable to capture execution history and unresolved research queries.

Compressing project context into a manifest reduces repeated scans when researchers switch contexts or AI agents take over tasks. It provides a ready-to-use, machine-readable workspace memory that allows future AI sessions to orient themselves instantly without rescanning the whole repo.