agentleFS
Data-WiseData-Wise/craft/skills/workflow/adhd-workflow/SKILL.md

adhd-workflow

Skill4 starsChanged 12 days agoMITView on GitHub

ADHD-friendly session and task workflow for craft users — captures context at session end, restores it at session start, picks the next task, breaks through blockers, holds focus on one task, reviews a craft spec, or refines a vague craft prompt. Use when wrapping up or resuming a coding session, when deciding what to work on next (also "what's next", "pick a task for me", "give me something to do"), when stuck or in decision paralysis, when focusing on a single craft task, or when reviewing/refining a craft spec or prompt. The full intent-to-operation map is in the "When to Use" table below.

The file

We split this skill into the rules it gives an agent

4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 2

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Distinctive:

    If multiple intents apply, pick the operation that matches the strongest verb in the prompt.

    line 24
  2. Very specific:

    Update .STATUS file: "✅ Just Completed", "🎯 Next Action", "🔴 Blockers".

    line 46
  3. Distinctive:

    Commit to one task and block mental distractions.

    line 82
  4. Very specific:

    Always read .STATUS first when restoring context.

    line 148

≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

ADHD Workflow Skill

Expert in ADHD-friendly session and task workflow for craft users. Reduces cognitive overhead at the points where ADHD costs the most: session boundaries (context loss), task selection (decision paralysis), being stuck (overwhelm), and unclear prompts (rework cost).

When to Use

Activate this skill when the user's prompt matches any of these concerns:

User intentOperation
"I'm done", "wrap up", "end session", "save progress"Session completion
"where did I leave off", "what's the state", "recap me"Context restoration
"what next", "what's next", "what should I work on", "I can't decide", "pick a task for me", "give me something to do", "what should I do now"Next-task suggestion
"let me focus", "single task", "Pomodoro"Focus mode
"I'm stuck", "I don't know what to do", "blocked"Unblock helper
"review my spec", "approve spec", "archive spec"Spec review
"fix my prompt", "rewrite this prompt", "optimize prompt"Prompt refinement

If multiple intents apply, pick the operation that matches the strongest verb in the prompt. If still ambiguous, default to recap (cheapest) and offer follow-ups.

Operations

1. Session Completion (done)

Capture progress before context evaporates — typically the highest-leverage moment in an ADHD work session.

Canonical procedure: the full session-completion flow lives in references/done.md — the single source of truth shared with the /craft:finish slash shim. For anything beyond the quick summary below (Settings Sync, Memory Optimize, Insights Capture, Worktree Status, the interactive summary, auto-git), load references/done.md and follow it. Never reimplement that flow here or in the command shim. See ADR-002.

Inputs: git status, git log --since="4 hours ago", current .STATUS file.

Quick summary (full detail in references/done.md):

  1. Gather session activity — uncommitted changes, recent commits, files modified.
  2. Summarize what was accomplished in 3–5 bullets, plain language.
  3. Update .STATUS file: "✅ Just Completed", "🎯 Next Action", "🔴 Blockers".
  4. Sync CLAUDE.md, audit settings drift, capture + optimize memory, write the insights facet (see reference for the exact steps).
  5. Suggest a conventional-commits message; surface follow-ups (open PRs, failing tests, half-written specs).

Producer role: This is the upstream data producer for /recap, /next, and /craft:hub — make .STATUS writes accurate and complete.

2. Context Restoration (recap)

Quickly answer "where did I leave off?" — the opener for any returning session.

Sources (priority order):

  1. .STATUS file (most authoritative) — "✅ Just Completed", "🎯 Next Action", "🔴 Blockers" sections.
  2. Recent git activity (last 48h) — git log --oneline --since="48 hours ago".
  3. Open PRs / issues via gh pr list --author @me --state open and gh issue list --assignee @me. When issues are returned, suggest /craft:git:issue-check <N> to check whether each issue's premise still holds before resuming work on it — not a broader GitHub-attention scan (no such skill exists; v1 scope is issue-premise-check only).
  4. Project planning files — TODO.md, PLAN.md, ROADMAP.md, CLAUDE.md.
  5. Obsidian sync config — run obs doctor --layer flow --json (if .flow/ exists). Report flow-sync-missing as ⚠️ + suggest obs flow init; report flow-sync-stale as ⚠️ + suggest obs flow init to update. Never block on flow warnings — they are informational.

Output: Short, scannable summary — what was completed, what's next, what's blocked. Aim for under 20 lines.

3. Next-Task Suggestion (next)

Cut decision paralysis by suggesting one clear next task — not a menu of five.

Priority order:

  1. Unblocked items (was waiting, now ready).
  2. In-progress work (maintain momentum).
  3. Quick wins (< 15 min, builds confidence).
  4. Important but not urgent.
  5. Blocked items — acknowledge but don't suggest.

Output: One recommendation with a one-sentence rationale. Offer at most one alternative if there's a close second.

4. Focus Mode (focus)

Commit to one task and block mental distractions.

Steps:

  1. Confirm the task — accept /next's suggestion, or take a user-supplied task description.
  2. Set focus parameters — suggested duration (25 min Pomodoro / 45 min standard / 90 min deep work).
  3. Display a focus banner with the task, time budget, and "what to ignore" reminders.
  4. Capture distractions to a list ("come back to these later") rather than acting on them.

5. Unblock Helper (stuck)

Targeted help when the user can't move forward.

Six common stuck types — ask which applies, then route:

TypeDiagnosticRouting
😵 Don't know where to startGoal unclear or task too bigSmallest-first-step breakdown
🤔 Don't understand somethingKnowledge gapExplain, link docs, or dispatch an Explore agent
🔧 Technical problem/errorBug or env issueSuggest /craft:code:debug or /bug-detective
😰 Overwhelmed by scopeTask too bigBreak into 15-min subtasks
🧱 Waiting on something externalTrue blockerAcknowledge, suggest parallel work
😴 Can't focus todayEnergy issueSuggest break, quick win, or non-coding task

6. Spec Review (spec-review)

Interactive review of formal specs in docs/specs/SPEC-*.md.

Sub-actions:

ActionBehavior
list (default if no topic)Enumerate specs with status
show <topic>Display spec content
review <topic>Validate frontmatter, walk acceptance criteria, prompt for status updates
approve <topic>Mark Status: approved
archive <topic>Move to docs/specs/archive/ (completed work)

Validation checks on review:

  • Frontmatter parses (Status:, Created:, Related plan: fields).
  • Acceptance criteria are testable (each maps to a check).
  • Open questions don't block the current batch.
  • History table is up to date.

7. Prompt Refinement (refine)

Rewrite vague prompts into specific, well-structured requests.

Steps:

  1. Get the prompt — argument or interactive.
  2. Diagnose — what's missing? (goal, constraints, success criteria, context, format).
  3. Rewrite — produce a clearer version that names the goal, constraints, expected output, and any project context the user implied.
  4. Explain — short bullet list of what changed and why, so the user learns the pattern.

Anti-patterns to fix: "help me with X" (no goal), "make it better" (no criteria), "fix this" (no context), missing file paths, missing acceptance criteria.

Cross-Operation Data Flow

This skill participates in a producer-consumer-reflector pattern with the rest of craft:

  • Producer: done writes .STATUS, commits, memory entries.
  • Consumer: recap, next, focus, stuck read .STATUS and session history before acting.
  • Reflector: /craft:hub aggregates the state into a dashboard.

Always read .STATUS first when restoring context. Always update .STATUS last when completing a session.

ADHD-Friendly Output Principles

  • One decision at a time — never offer 5 options when 1 will do.
  • Visual hierarchy — headers, tables, emoji markers (🎯 ✅ 🔴 🚧) for scannability.
  • Concrete next step — every operation ends with a clear, actionable suggestion.
  • Quick wins highlighted — explicitly label items under 15 minutes.
  • No "while you're here" — stay scoped to the operation requested; capture-and-defer everything else.

Integration

This skill replaces the seven commands/workflow/*.md commands during the v2.34.0 → v3.0.0 migration:

  • /done → operation 1 (Session Completion)
  • /recap → operation 2 (Context Restoration)
  • /next → operation 3 (Next-Task Suggestion)
  • /focus → operation 4 (Focus Mode)
  • /stuck → operation 5 (Unblock Helper)
  • /spec-review → operation 6 (Spec Review)
  • /refine → operation 7 (Prompt Refinement)

Both invocation paths work during the deprecation cycle. The skill auto-fires on natural-language match; explicit slash paths continue to function until v3.0.0 — /craft:finish (renamed from /craft:done, ADR-006), /craft:next, and /craft:refine are now root commands (promoted from workflow:), while /craft:workflow:recap, /craft:workflow:focus, and /craft:workflow:spec-review remain nested.

For ADHD-friendly workflow strategy prose (not an operation, no behavior), see references/adhd-guide.md.

Related Skills

  • mode-controller — switches execution mode (default/debug/optimize/release) orthogonally to these operations.
  • project-planner — for multi-week project breakdowns (this skill handles session-scope; project-planner handles project-scope).
  • release — when the next action is "ship it".

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.