--- name: things-week-ahead-digest description: Summarize Things todos and recommend next actions using recent activity across projects, areas, and checklist-like notes, plus a week/weekend-ahead preview. Use when asked for Things planning digests, productivity check-ins, weekly previews, or automation that turns Things data into concise priorities. --- # Things Week Ahead Digest Build a repeatable Things planning digest with four sections: - Snapshot - Recently Active - Week/Weekend Ahead - Suggestions ## Inputs Collect data with Things MCP tools: 1. `things_read_areas` 2. `things_read_projects` with `status="open"` 3. `things_read_todos` with `status="open"` (set `limit` high enough for a full view, usually 300-500) 4. `things_read_todos` with `status="completed"` and `completed_after=` 5. Optional: `things_read_todo` for top candidate tasks (`include_notes=true`) when notes/checklist signals are needed If any call fails because of permissions, report the missing permission and continue with the best available subset. ## Workflow 1. Load active customization config: - Prefer `/config/customization.yaml`. - Fall back to `/config/customization.template.yaml`. 2. Build urgency buckets from open todos: - overdue: `deadline < today` - due soon: `deadline <= today + ` - week/weekend ahead: `deadline` between today and `today + ` 3. Score recent activity by project and area using configured weights: - recent completions (last 7 days) - open tasks - due soon / overdue counts - checklist-like hints from notes (`- [ ]`, `- [x]`, multiline bullet blocks) 4. Identify top active projects/areas from the score. 5. Generate configured number of concrete suggestions: - one next action for top projects - one risk/triage action for overdue work - one planning action for weekend or Monday readiness when relevant 6. Format with the template in `references/output-format.md`. Keep tone concise and operational. Prefer verbs and specific task titles over generic advice. ## Scripts Use `scripts/build_digest.py` when deterministic scoring/formatting is preferred (automation, repeatability, large datasets). ```bash uv run --with pyyaml python scripts/build_digest.py \ --areas areas.json \ --projects projects.json \ --open-todos open_todos.json \ --recent-done recent_done.json ``` The script reads JSON exported from Things MCP responses and prints Markdown digest output. Configuration precedence: 1. CLI flags 2. `config/customization.yaml` 3. `config/customization.template.yaml` 4. Script hardcoded defaults ## Customization Workflow When a user asks to customize this skill, use this deterministic flow: 1. Read active config from `config/customization.yaml`; if missing, use `config/customization.template.yaml`. 2. Confirm desired behavior for: - due-soon and planning windows - top project/area counts - scoring weights and open-count cap - suggestion cap and output style 3. Propose 2-4 option bundles with one recommended default. 4. Create or update `config/customization.yaml` from template and set: - `schemaVersion: 1` - `isCustomized: true` - `profile: ` 5. Validate by generating a sample digest and report changed keys plus behavior deltas. ## Customization Reference - Detailed knobs and examples: `references/customization.md` - YAML schema and allowed values: `references/config-schema.md` ## Automation Templates Use `$things-week-ahead-digest` inside automation prompts so Codex reliably loads this workflow. For ready-to-fill Codex App and Codex CLI (`codex exec`) templates, including Things MCP fallback guidance, placeholders, and customization knobs, use: - `references/automation-prompts.md` ## References - Scoring and suggestion rules: `references/suggestion-rules.md` - Output shape and section template: `references/output-format.md` - Automation prompt templates: `references/automation-prompts.md` - Customization guide: `references/customization.md` - Customization schema: `references/config-schema.md`