How we plan and track work: skeleton

Target: a short guide a new team member, or an AI, can read to get up to speed on how Life Itself plans and tracks work now. Draft at 2026-10-22-planning-guide-draft.md. Method per writing briefings and guides: skeleton and sketch, critique, draft on a separate page, critique, Rufus review.

Sources

• Decisions: PM infrastructure LTP report (tools cloud, build vs ship, future reality check with flow vs state and the project catalogue, prerequisite tree, task routing test) and the briefing.
• Current rules, partly out of date: workflow, README, data model, Beads conventions, planning skill, procedures.
• Neighbouring guides: Our office, Beads, Way Into AI's Post-its and filing cabinets and AI is the interface, Tao's Getting Stuff Done and onboarding.
• Live state checked in Beads (read only) on 22 Oct: lip-tqk.23 (line, open), .26.1 (slot invite), .30 (crossover rule, due 29 Oct), .34 (weekly plan format), .37 (task convention), .47 (trial dates), .48/.49 (Thursday plan and handover), .54 (meeting notes); shared skills home decided as life-itself/skills (.29.1). No initiative record has a priority: field yet.

Audience and stance

• Reader: a new team member who has read (or will read) Post-its and filing cabinets, and an AI agent opening the planning repo. Neither knows our history.
• One message: records hold the current truth (state); tasks sit in one tracker per kind of work; once a week Rufus and Uriel turn both into a plan; the AI builds your view and links its source.
• How-to more than concept guide, so it keeps a short opening (one line, the ask, the status) rather than a full TL;DR, then the picture. Concepts first, then where things live, then the weekly rhythm, recipes, and tools last.
• It does not repeat Our office (the repo-by-repo map) or the Beads guide; it links them.
• As of 22 Oct the tools split is a 2 to 4 week trial. Every routing rule is marked decided or to agree, so the open ones can be settled and the marks removed.

Concept map and order

Orders tried:

1. Top down (initiative, project, task): matches the README tree, but starts with the most abstract word.
2. Bottom up (task, project, initiative): smallest first, as the concepts-guide rule says, but a lone task has no context and the reader asks "for what?".
3. One project, then zoom out and in (chosen): start from a project the reader can picture (the Global Connection Call, as in Our office), zoom out to its initiative and area, zoom in to its tasks and papers, then state vs flow using its priority. A project is obviously one piece of work with an end, and the example ties every concept to one thing.

The one picture (sketch)

Tall and narrow; argument-diagram style (numbered steps, gist, icon, arrows, key idea marked, the clean-up loop as a branch).

 HOW WE PLAN AND TRACK WORK

 ① [▤] RECORDS ─ what and why
       initiative ▸ project
       owner · status · priority
       Markdown in Git
       = STATE   ◀ the key idea:
         the current truth lives here
          │ each record links its tracker
          ▼
 ② [▦] TASKS ─ what to do next
       owner · next action · priority
       Media House     → Asana
       ops + planning  → Beads / GitHub
       crossover       → one owner, others link
          │ the AI sweep reads them
          ▼
 ③ [◷] WEEKLY PLAN ─ Thursday 12–2
       Rufus + Uriel (Asdra learning)
       focus · owners · task IDs
       = FLOW: history, not current priority
          │ priority changes go back to ①
          ▼
 ④ [◉] YOUR VIEW ─ "what's mine this week?"
       Claude + planning skill
       links its source → check it
          │
          └─▷ ⑤ clean-up: stale, unowned,
                duplicate → a person confirms
                → fix in ① and ②

Skeleton

Opening (5 lines + picture)

• One line: how we plan and track work at Life Itself, for new team members and the AI.
• Ask: nothing; read once, then use the recipes. For the 22 Oct meeting: settle the rows marked "to agree".
• Status: as of 22 Oct 2026; the tools split is in a trial led by Uriel; rows marked "to agree" are still open.
• The picture above.

1. One project, four kinds of thing (~250 words + record sketch)

• Project: bounded work with an outcome and an end. Example: the Global Connection Call, done when the first call series has run.
• Initiative and initiative area: the enduring areas it sits in (Calls and Catalysts, within Gardens of Change, within Seeds of Renaissance). Areas nest; each has a record.
• Tasks: the to-dos ("book the first speaker"), kept in a tracker, never as lists in documents.
• Reference material: the papers (why, the call pattern, notes per call): Markdown in a Git repo.
• Small sketch of the project record, one point per line (what, owner, status, priority, done when, tracker, papers).

2. State and flow (~150 words + small picture)

• Bank analogy: the balance (state) vs the transactions (flow).
• State: current priority and status live on initiative and project records. Initiatives carry priority set by Rufus and Uriel; projects carry priority relative within their initiative, set by the initiative owner.
• Flow: weekly plans and meeting notes record what happened. Read them for history; nobody should have to read old plans to know what matters now.
• Note: priorities are agreed but not yet written onto the records.

3. Where things live (~200 words + routing table)

• One system of record per kind of work; link, don't copy.
• Table columns: what, where, status (decided / to agree). Rows: initiatives and areas (planning repo initiatives/, decided); portfolio index listing every initiative and area with its repo (planning, decided, being completed); projects (projects/ folder in the initiative's repo, or a numbered record in planning; decided); reference material (Markdown in Git on GitHub, moved from Drive as touched; decided); Media House / content production tasks (Asana, decided in principle); operations, infrastructure, code and planning tasks (Beads and GitHub, decided in principle); exact Asana line (to agree; Uriel's proposal); Beads vs GitHub Issues (to agree); which repo's Beads (to agree; current practice stated); crossover work (decided: one owning record, others link; rule being written); meeting notes (to agree); shared skills (life-itself/skills, decided).
• Short note on the old rule in workflow.md being replaced.

4. The weekly cycle (~200 words + loop picture)

• Thursday 12:00–14:00: weekly planning and system work in one protected slot; delivery can slip to protect it.
• Owners: Rufus and Uriel; Asdra apprenticing as Uriel's deputy. Uriel is air traffic controller for incoming requests (Sylvie's included).
• Steps: run the sweep; draft the team plan in the Google Doc tab; agree; hand the week's tasks to Amanda, Asdra and Fae with owner, next action, priority; copy to weeks/YYYY-MM-DD.md (dated Monday).
• Priority changes: Rufus and Uriel change them on the records; the plan only mentions them. Convention changes are briefed in the slot.
• Format still a pilot.

5. Recipes (~400 words, five numbered lists)

• Start a project (check portfolio; record in initiative repo's projects/ or in planning; minimal fields; link tracker; project-planning skill for big or new work).
• Log a task (find the right tracker from the table; owner at epic level, next action, priority; link the project; check for duplicates).
• Hand work across teams (one owning record; others link; requests to Uriel; no copying).
• Ask the AI what to do this week (prompt to paste; check the source it links; what it can't read yet: Asana).
• Clean up (sweep; close finished with a reason; merge duplicates; add owners; a person confirms; Asdra's data steward routine).

6. Tools and skills (~200 words + short table)

• Table: tool, used for, who, how to get it. Claude desktop (Code tab); planning skill (sweep, week); portfolio skill; project-planning skill; Beads (bd); GitHub (gh); Asana (Media House).
• Install: Tao onboarding "Set up your computer (with AI)"; skills from life-itself/skills; set BEADS_ACTOR.

7. Read next, and who to ask (~100 words)

• Reading path: Post-its and filing cabinets → Our office → this guide → AI is the interface → Beads.
• Ask: Rufus (planning repo, Beads, records); Uriel (Asana, Media House, priorities and requests); Asdra (clean-up, weekly plan); Fae (planning site for collaborators).

8. Still being agreed (as of 22 Oct) (~120 words)

• The open rows from the table plus: trial dates, priority scales across tools, weekly copying of requests into Asana, weekly plan format, task convention for Asana. Each with owner and Beads ID.

More detail

• Links: LTP report, decisions briefing, workflow, data model, Beads conventions, planning skill, Our office, Beads guide.

Where should the guide live?

Overlap with Tao: Getting Stuff Done is now a reading path (Post-its, Our office, AI is the interface, setup, Beads) plus a three-line "Where things live"; it holds no rules. Issues covers writing GitHub issues and already defers to the planning workflow for tracker choice. Onboarding points to Planning for "Issues and Tasks". So the overlap is small: the routing rule and the task convention, both of which Tao already delegates to planning.

Recommendation: publish in the planning repo as guides/how-we-plan.md (order 2, after Our office) and make docs/workflow.md a slim reference behind it; add it as a step in Tao's Getting Stuff Done path rather than putting it in Tao. Reason: the rules change weekly during the trial and sit next to the records, skills and Beads they describe; Tao keeps culture and the reading path, and already points here for planning.

Open questions for the skeleton

1. "Initiative area": the LTP uses it without defining it. Proposed reading: an initiative nested inside a larger one (Podcast & YouTube within Media House), with its own record and often its own repo. Recommendation: use that and drop the term if Rufus prefers just "initiative".
2. Should "Still being agreed" be a section or only marks in the table? Recommendation: both for now (the section is today's agenda), and delete the section once settled.

Critique applied (22 Oct)

A separate agent reviewed this skeleton. Applied in the draft:

• The picture now starts from the project (matching the text), shows its papers as a branch separate from the record, and draws the state/flow loop as a back-arrow from the weekly plan to the record.
• "Initiative area" kept only as a one-line aside ("initiatives nest"); the data model has no separate type.
• Projects row adds the A10 Google Doc option and marks the projects/ folder convention as decided but not yet written; the Global Connection Call is shown as it is (record here, papers in the community repo), not as the model layout.
• Slot marked decided, with Rufus as slot owner, distinct from the weekly-plan owners (Rufus and Uriel).
• Status line uses the latest source: the trial started the week of 19 Oct (2605 record, LTP report), review due 3 Dec.
• Meeting agenda line removed from the guide; the ask is "nothing".
• "Ask the AI" recipe gives the actual prompt and what the AI can't read yet.
• Not applied: cutting to about 1,000 words, dropping "Still being agreed" and folding "hand work across teams" into "log a task". The brief asks for 1,200 to 1,800 words and for these items; the "Still being agreed" section stays short and goes once the trial review settles it.