| name | phase-output |
| description | Phase guidance for the neuroflow /output command. Orients agent approach for outputting project memory or the whole project safely and with correct exclusions. |
phase-output
The /output command packages and moves project data out of the workspace. Its job is to give the user a clean, portable snapshot without accidentally including sensitive files.
Approach
- Clarify the user's intent before choosing scope — "sharing with a collaborator" suggests memory-only; "handing off the full repo" suggests whole project
- Always exclude
sessions/ (local-only, can be large) and integrations.json (API credentials — must never be shared)
- Prefer zip over folder copy for sharing — it is a single file and preserves timestamps
- Use Python's
zipfile module first; fall back to system zip/tar only if Python is unavailable
- If the user asks about Notion integration or other cloud destinations, acknowledge it is planned but not yet implemented — offer zip as the portable alternative for now
- Do not modify the project or
.neuroflow/ state during export — the command is read-only except for writing the export log
Scope guidance
| Scope | Best for |
|---|
Project memory (.neuroflow/ minus exclusions) | Sharing context with a collaborator who has their own codebase; supervision meetings; archiving project decisions and reasoning |
Whole project (git-tracked + .neuroflow/) | Full handoff, long-term archiving, submitting to a data repository |
| Single phase | Focused handoff — e.g. sending only the data-analyze/ memory to a statistician |
Exclusions — always enforce
| Excluded | Reason |
|---|
.neuroflow/sessions/ | Local-only operation log — large, personal, not meaningful to external recipients |
.neuroflow/integrations.json | API credentials (PubMed email, Miro token) — must never leave the local machine |
If the user explicitly asks to include sessions or credentials, explain why that is inadvisable and confirm they still want to before proceeding.
File naming convention
Default names for export outputs:
output-[project-slug]-[YYYY-MM-DD].zip
output-[project-slug]-[YYYY-MM-DD]/ (folder copy)
output-[phase]-[project-slug]-[YYYY-MM-DD].zip (single-phase)
Where project-slug is the project name from project_config.md, lowercased with spaces replaced by hyphens.
What to suggest if the user asks "what else can I export?"
- Manuscript draft — if
paper/ exists, offer to run /write-report to generate a summary first, then include it in the export alongside any manuscript files from the paper output path
- Analysis report — run
/write-report first to generate a summary, then include it in the export
- Reasoning log — mention that
.neuroflow/reasoning/ is included in the memory export and contains all documented decisions
Do not suggest Notion or other cloud integrations — they are not implemented yet.
Relevant skills
neuroflow:neuroflow-core — read first; defines the command lifecycle and .neuroflow/ write rules
neuroflow:notebooklm — when the user wants a podcast, infographic, slide deck, or audio summary generated from exported project materials, invoke the notebooklm skill to create a NotebookLM notebook and feed the exported files into it as sources