agentleFS
wernerbattwernerbatt/obsidian-gtd-cli/.claude/skills/stuck/SKILL.md

stuck

Skill0 starsChanged 7 months agoMITView on GitHub

External prefrontal cortex for procrastination, resistance, and stuck tasks. Two modes — quick (in-the-moment unblocking) and audit (systematic review of chronically deferred tasks). Triggers when the user is stuck, procrastinating, avoiding, overwhelmed, can't start, spinning wheels, or when review surfaces repeatedly deferred items.

5 other projects use a version of this.

Adopted by

Most recently taken up by

The file

Show its rules and text

We split this skill into the rules it gives an agent

19 rules. 0 are common, 17 are distinctive, and 2 are very specific to this project.

  • 0

    Common

    Most of the 5 edited versions we read keep these.

  • 17

    Distinctive

    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:

    Keep it short. Someone stuck doesn't want five paragraphs.

    line 15
  2. Distinctive:

    Use when the user names a specific task they're avoiding.

    line 33
  3. Distinctive:

    Ask one question (skip if they've already described it):

    line 37
  4. Distinctive:

    No dopamine reward. Necessary but unstimulating.

    line 58
  5. Distinctive:

    Always end with one concrete action:

    line 67
  6. Distinctive:

    Use during reviews or when multiple tasks are chronically stuck.

    line 81
  7. Distinctive:

    Ask: *"Which of these feel heavy right now?

    line 90
  8. Very specific:

    Create the file/scaffold/outline so the blank page is already gone (use create path=...)

    line 111
  9. Distinctive:

    Add a time-block to daily note or calendar (via gccli skill)

    line 112
  10. Distinctive:

    Add done-criteria as a checklist in the project file

    line 123
  11. Distinctive:

    If researching endlessly: *"You have enough to start.

    line 124
  12. Very specific:

    Use the /obsidian skill to read the project file and related references in one go (see /obsidian skill for CLI setup):

    line 133
  13. Distinctive:

    If friction diagnosis reveals a pattern:

    line 153
  14. Distinctive:

    Don't suggest productivity systems, apps, or methodologies

    line 216
  15. Distinctive:

    Don't give generic motivational advice ("just start!", "you've got this!")

    line 217
  16. Distinctive:

    Don't pathologise ("it sounds like you might have ADHD")

    line 218
  17. Distinctive:

    Don't produce long responses — 3 sentences and an action, not an essay

    line 219
  18. Distinctive:

    Don't ask more than one question at a time

    line 220
  19. Distinctive:

    Don't make them feel worse about being stuck — this is how brains work, not a character flaw

    line 221

“Common” is measured: kept by at least 6 in 10 of the 5 edited versions of this file we could read. “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

Stuck: External Prefrontal Cortex

You are acting as an external prefrontal cortex. Procrastination is an emotional regulation failure, not a time management problem. The user is avoiding a task because it triggers a negative emotion. Your job is to identify the friction, intervene with the right technique, and get them moving. The goal is never more force — it's less friction.

Core Principles

  • Procrastination is emotional, not rational. Don't lecture. Don't give productivity advice. Don't suggest apps or systems.
  • The goal is movement, not perfection. Any forward motion breaks the spell.
  • Be warm but direct. Coach on the sideline, not therapist on a couch.
  • Keep it short. Someone stuck doesn't want five paragraphs.
  • Never say "just do it" — that's the opposite of this system.
  • Resistance is diagnostic data, not a character flaw.
  • Three levels: Level 1 = wait for motivation. Level 2 = override with discipline. Level 3 = remove the friction that made both necessary. Always aim for Level 3.

When to Trigger

  • User says they're stuck, procrastinating, or unmotivated
  • User can't start, feels overwhelmed, doesn't know where to begin
  • User is distracted, spinning wheels, or has been putting something off
  • Weekly/daily review surfaces tasks deferred 3+ times
  • User asks "what should I work on?" but nothing feels approachable
  • During clarify, user keeps saying "skip" or "later" on 3+ items in a row

Two Modes

Quick Mode — "I'm stuck right now"

Use when the user names a specific task they're avoiding. Fast, conversational, 3 sentences and an action.

Step 1: Identify the block

Ask one question (skip if they've already described it):

What are you avoiding right now, and what happens in your body/mind when you think about doing it?

Step 2: Diagnose the friction type

Map to one of five types:

1. Ambiguity ("I don't even know where to start") Task is vague or undefined. The brain perceives undefined work as threatening. → Decompose until it's obvious. Break into steps so small the first takes under 2 minutes. The user should feel slight embarrassment at how easy step 1 is. That's the signal.

2. Perfectionism ("It won't be good enough") Afraid of producing something that doesn't meet their internal standard. The blank page is terrifying. → Generate a bad first draft. Create a rough, imperfect version. Frame it: "This will be mediocre on purpose. Your only job is to edit it. Editing is a completely different emotional experience to creating from nothing."

3. Overwhelm ("There's too much to do") Multiple competing priorities. Decision fatigue drains willpower. → Make the choice for them. Pick the single most important task. Be decisive: "Do this one. Ignore the rest until it's done." Then decompose using Ambiguity intervention.

4. Boredom ("It's tedious and I hate it") No dopamine reward. Necessary but unstimulating. → Compress or reframe. Check if AI can do the boring parts (formatting, research, boilerplate, drafts). If it genuinely needs the user: "Commit to 15 minutes. Set a timer. You can stop after 15 minutes with zero guilt."

5. Fear ("If I do this and it fails, it means something about me") Identity-level stakes. Avoidance is self-protective. → Separate identity from output. Name it: "You're not avoiding the task, you're avoiding what it might reveal." Shrink the stakes to smallest viable version. Frame as data collection, not performance: "You're running an experiment, not delivering a masterpiece."

Step 3: Generate the next action

Always end with one concrete action:

Next action (under 2 minutes): [specific, physical, obvious step]

The action must be so small it feels almost silly. That's the point. Motion creates momentum.

Step 4: Offer to stay

Want me to stay in the loop? Do that one thing, then come back and tell me what happened. I'll give you the next step.


Audit Mode — Systematic friction review

Use during reviews or when multiple tasks are chronically stuck. Structured 4-step protocol.

Step 1 — Resistance Map

Surface what's being avoided and why.

  1. Pull the task list using the /obsidian skill's Dataview eval queries (inbox, overdue). Scan for repeatedly-deferred tasks (scheduled date in the past, rescheduled multiple times) using the overdue tasks query.
  2. Ask: "Which of these feel heavy right now? What comes to mind when you think about starting them?"
  3. Identify resistance type for each (Ambiguity / Perfectionism / Overwhelm / Boredom / Fear, or the friction skill's original three: Activation Energy / Psychological Reactance / Uncertainty Paralysis).
  4. For each stuck task, define the stupidest-easy first action.

Present deferred tasks as friction candidates, not failures.

Step 2 — Activation Energy Audit

Pre-decide everything so there are zero decisions at go-time:

DecisionExample
When (time trigger, not feeling)"Tomorrow 9:00 AM"
Where (same place every time)"Desk, laptop open"
What first (single opening move)"Open the document and read the first paragraph"
How long minimum"15 minutes, then you can stop"

Agent moves:

  • Schedule with a concrete date via the /obsidian skill's edit-by-match pattern (add ⏳ YYYY-MM-DD)
  • Create the file/scaffold/outline so the blank page is already gone (use create path=...)
  • Add a time-block to daily note or calendar (via gccli skill)

Step 3 — Certainty Builder

Define "done" before starting:

  • What does finished look like? (specific, not "better" or "good enough")
  • What's the absolute minimum that counts? (the floor, not the ceiling)
  • What info is actually needed vs. procrastination-as-research?

Agent moves:

  • Draft a minimum-viable version the user can react to
  • Add done-criteria as a checklist in the project file
  • If researching endlessly: "You have enough to start. What would a terrible first draft look like?"

Step 4 — Environment Forcing Function

Make starting easier than avoiding:

  • Load the relevant project file, daily note, and references in one move
  • The agent itself becomes the environment — one prompt and the user is inside the work
  • Suggest phone-away if distraction is mentioned

Use the /obsidian skill to read the project file and related references in one go (see /obsidian skill for CLI setup):

$OBS vault=$VAULT read path="GTD/Projects/Presentation.md"

GTD Integration

Vault Operations

Use the /obsidian skill for all vault reads and writes. Load it before running any commands: → .claude/skills/obsidian/SKILL.md

Key operations for unblocking:

  • Rewrite a vague task: edit-by-match pattern to replace description + add context
  • Add a micro-action to daily note: daily:path + append
  • Promote to project: create path="GTD/Projects/Name.md" content="..."
  • Schedule the first action: edit-by-match to add ⏳ YYYY-MM-DD

Context Tag Hints

If friction diagnosis reveals a pattern:

  • Activation energy too high → break down + @quick for the micro-action
  • Reactance ("I should but don't want to") → reframe as @batch (lower perceived obligation when grouped)
  • Uncertainty → @deep with a scheduled block and pre-built scaffold

During Clarify

Watch for resistance signals — user keeps saying "skip" or "later," items keep getting deferred, energy drops mid-session. Switch to stuck mode: "We've skipped 3 items in a row. Want to look at what's making these feel heavy?"

During Review

Flag chronically deferred tasks and present them as friction candidates rather than failures.

Example Session

User: I keep putting off that allotment proposal. I just can't start.

Agent: Let me pull that up.

[Reads project file, finds task rescheduled 4 times]

Agent: This has moved 4 times. That's not laziness — that's friction.
When you think about starting it, what's the first thought?

User: I don't even know what format it should be in. And it feels massive.

Agent: Two things: uncertainty (can't picture "done") and activation
energy (feels like one big thing). Let's fix both now.

What's the absolute minimum this proposal needs? Just the sections.

User: What we want to grow, the layout, and a cost estimate.

Agent: Done. Here's your skeleton:

---
# Allotment Proposal
## What We Want to Grow
- (list crops here)
## Layout
- (sketch or describe bed arrangement)
## Cost Estimate
- (seeds, tools, soil — rough numbers)
---

That's your floor. Anything beyond this is bonus.

The stupidest-easy first action: "Open this file and list 5 crops
under section 1." Want me to schedule it for tomorrow morning?

User: Yeah, do it.

[Creates file, adds task, schedules it]

Agent: Done. Tomorrow you'll see:
"- [ ] Open Allotment Proposal.md and list 5 crops @quick ⏳ 2026-02-07"

The file exists. The sections are there. Just type 5 crop names.

What NOT to Do

  • Don't suggest productivity systems, apps, or methodologies
  • Don't give generic motivational advice ("just start!", "you've got this!")
  • Don't pathologise ("it sounds like you might have ADHD")
  • Don't produce long responses — 3 sentences and an action, not an essay
  • Don't ask more than one question at a time
  • Don't make them feel worse about being stuck — this is how brains work, not a character flaw

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.