michaelalber/para-file

PARA File

"Organize by actionability, not by subject. The right question is not what is this about? but which project will this move forward?" — adapted from Tiago Forte, fortelabs.com/blog/para

Core Philosophy

PARA sorts every piece of information into exactly four categories that exist identically on every platform: Projects (short-term efforts with a goal and an end), Areas (ongoing responsibilities with a standard to maintain, no end date), Resources (topics of ongoing interest / reference), and Archives (inactive items from the other three). The discipline that makes PARA work is sorting by actionability, not topic — a document about "marketing" is not filed under a "Marketing" folder; it is filed wherever it is most actionable right now.

This skill files one item at a time. It is deliberately fast — the cost of a slightly-wrong filing is near zero because items move freely between categories as their status changes, and para-review catches drift later. Bias toward placing the item and moving on, not toward a perfect taxonomy.

Non-Negotiable Constraints:

1. ACTIONABILITY ORDER — evaluate Projects → Areas → Resources → Archives, in that order, and file at the first match. The most actionable home wins. (Decision tree in references/classification-heuristics.md.)
2. FOUR CATEGORIES ONLY — never invent a fifth top-level bucket. Sub-structure lives inside P/A/R.
3. NEVER DESTROY — filing only creates or moves; it never deletes. Overwrites require explicit confirmation.
4. CONFIG-DRIVEN — read .para.yml for every root, label, and backend. If absent, scaffold and write it.
5. CONFIRM CROSS-BACKEND — publishing to Confluence/Jira is outward-facing; confirm before the first write in a session.

The full category criteria, edge cases, and anti-patterns live in references/classification-heuristics.md.

Workflow

DETECT      Read .para.yml (project root). Resolve local roots, OneDrive/Teams synced paths,
            Confluence space + label scheme, Jira active-project keys. If config is missing OR a
            declared local root has no Projects/Areas/Resources/Archive tree → SCAFFOLD.
            No files moved yet. (Schema: references/config-schema.md.)

SCAFFOLD    (only if missing) Create the four roots + an inbox under each declared local/OneDrive
            base; write a starter .para.yml. Confluence/Jira are mapped, never auto-created — report
            the label/space scheme for the user to confirm. Idempotent: never clobber an existing tree.

CLASSIFY    Apply the actionability decision tree to the item. Output: category (P/A/R/Archive),
            target location (folder | Confluence parent | Jira project), and a one-line rationale.
            If ambiguous between two categories, state both and pick the more actionable; do not stall.

SUMMARIZE   Produce a 1–3 sentence summary + 2–5 topic tags for the item, to aid retrieval and to
            confirm the classification. Skip only for items the user marks trivial.

FILE        Execute via the matching backend adapter (references/backend-adapters.md):
              local / OneDrive / Teams → move-or-create file into <root>/<category>/...
              Confluence              → create/label page under the mapped space
              Jira                    → tag/associate with the active project (Projects only)
            Capture the result path/URL. Never delete the source without confirmation.

RECORD      Append the filing (item, category, destination, tags, summary) to the PARA index note
            and emit the result. Update .para.yml only if a new root/label was introduced.

Exit criteria: item placed in exactly one category via the correct backend; summary + tags recorded; destination path/URL reported; no source destroyed; .para.yml consistent.

State Block

<para-file-state>
phase: DETECT | SCAFFOLD | CLASSIFY | SUMMARIZE | FILE | RECORD | COMPLETE
config_found: true | false
scaffolded: true | false | n/a
item: [name or description]
backend: local | onedrive | teams | confluence | jira
category: project | area | resource | archive | undecided
destination: [path or URL]
tags: [comma-separated]
source_preserved: true | false
last_action: [description]
next_action: [description]
</para-file-state>

Output Template

• .para.yml schema + scaffolding layout — references/config-schema.md.
• Category criteria, actionability decision tree, edge cases, anti-patterns — references/classification-heuristics.md.
• Per-backend file/move/label operations (filesystem, OneDrive/Teams, Confluence + Jira via Atlassian MCP) — references/backend-adapters.md.

Integration with Other Skills

| Skill | Relationship | |-------|-------------| | para-review | Run periodically after filing — audits structure, runs the weekly ritual, and archives completed items. | | confluence-guide-writer | Use when a filed Confluence page should be expanded into a full user guide. | | jira-comment-writer | Use when filing relates to an active Jira project and needs a structured comment. | | research-synthesis | Use when a filed Resource warrants deeper multi-source research before filing. |