Prompt file imported from PauloHYBEX/Nexus-TCMS (.windsurf/workflows/pause-work.md). Copyright stays with the author.

Pause Work

Name: pause-work
Author: PauloHYBEX

Create a .continue-here.md handoff file that captures complete work state. Enables seamless resumption with full context restoration.

Use when: stopping mid-plan, mid-phase, or any time you want to preserve exact position.

Step 1: Detect Current Phase

Find the most recently active phase:

node -e "const fs=require('fs'),path=require('path');function find(d){let r=[];try{for(const e of fs.readdirSync(d,{withFileTypes:true})){const f=path.join(d,e.name);r=r.concat(e.isDirectory()?find(f):e.name.endsWith('-PLAN.md')?[f]:[]);}}catch(e){}return r;}const files=find('.planning/phases').map(f=>({f,t:fs.statSync(f).mtimeMs})).sort((a,b)=>b.t-a.t);if(files[0])console.log(files[0].f);"

Extract the phase directory name from the result.

If no active phase detected, ask: "Which phase are you pausing work on?"

Step 2: Gather State

Collect the complete handoff state by reading current files:

cat ".planning/phases/[phase_dir]/"*-PLAN.md 2>/dev/null
cat .planning/STATE.md

Also ask the user (conversationally) for anything that can't be inferred from files:

Which specific task were you on?
What's already completed vs. what's in progress?
Any key decisions made this session that aren't yet committed?
Any blockers or things to watch out for?
What's the very first thing to do when resuming?

Step 3: Write Handoff File

Write .planning/phases/[phase_dir]/.continue-here.md:

---
phase: [phase_dir]
task: [current task number]
total_tasks: [total tasks in current plan]
status: in_progress
last_updated: [ISO timestamp]
---

<current_state>
[Where exactly are we? Phase X, Plan Y, Task Z — immediate context for a fresh session]
</current_state>

<completed_work>
- Task 1: [name] — Done
- Task 2: [name] — Done
- Task 3: [name] — In progress, [what's done so far]
</completed_work>

<remaining_work>
- Task 3: [what's left to complete]
- Task 4: Not started
- Task 5: Not started
</remaining_work>

<decisions_made>
- Decided to use [X] because [reason]
- Chose [approach] over [alternative] because [reason]
</decisions_made>

<blockers>
- [Any blocker]: [status or workaround]
</blockers>

<context>
[The mental model: what approach is being taken, why, what to watch out for]
</context>

<next_action>
Start with: [the specific first action to take when resuming — be precise]
</next_action>

Be specific enough that a fresh Cascade session can pick up immediately without re-asking.

Step 4: Commit

git add ".planning/phases/[phase_dir]/.continue-here.md"
git commit -m "wip: [phase-name] paused at task [X]/[Y]"

Step 5: Confirm

✓ Handoff created: .planning/phases/[phase_dir]/.continue-here.md

Current state:
- Phase: [phase_dir]
- Task: [X] of [Y]
- Status: in_progress
- Committed as WIP

▶ To resume: resume-work
▶ Capture an idea before you go: /note [text]
▶ Generate a session summary: /session-report

Learning Checkpoint

Read learning_mode from .planning/config.json.

If auto: Offer before the session ends:

💡 Before you go: Session transitions are when learning decays fastest. Two minutes now saves re-learning later:

@agentic-learning space — Identify what was covered this session and schedule it for spaced review. Writes to docs/revisit.md. When you resume, you'll start sharper.

@agentic-learning reflect — Quick 3-part reflection: what was built this session, what was the intent, what still feels uncertain. Takes 3 minutes, pays off when you return.

If manual: Add quietly: "Tip: @agentic-learning space before closing to schedule what you worked on for review."

# Pause Work Create a `.continue-here.md` handoff file that captures complete work state. Enables seamless resumption with full context restoration. **Use when:** stopping mid-plan, mid-phase, or any time you want to preserve exact position. ## Step 1: Detect Current Phase Find the most recently active phase: ```bash node -e "const fs=require('fs'),path=require('path');function find(d){let r=[];try{for(const e of fs.readdirSync(d,{withFileTypes:true})){const f=path.join(d,e.name);r=r.concat(e.isDirectory()?find(f):e.name.endsWith('-PLAN.md')?[f]:[]);}}catch(e){}return r;}const files=find('.planning/phases').map(f=>({f,t:fs.statSync(f).mtimeMs})).sort((a,b)=>b.t-a.t);if(files[0])console.log(files[0].f);" ``` Extract the phase directory name from the result. If no active phase detected, ask: "Which phase are you pausing work on?" ## Step 2: Gather State Collect the complete handoff state by reading current files: ```bash cat ".planning/phases/[phase_dir]/"*-PLAN.md 2>/dev/null cat .planning/STATE.md ``` Also ask the user (conversationally) for anything that can't be inferred from files: - Which specific task were you on? - What's already completed vs. what's in progress? - Any key decisions made this session that aren't yet committed? - Any blockers or things to watch out for? - What's the very first thing to do when resuming? ## Step 3: Write Handoff File Write `.planning/phases/[phase_dir]/.continue-here.md`: ```markdown --- phase: [phase_dir] task: [current task number] total_tasks: [total tasks in current plan] status: in_progress last_updated: [ISO timestamp] --- <current_state> [Where exactly are we? Phase X, Plan Y, Task Z — immediate context for a fresh session] </current_state> <completed_work> - Task 1: [name] — Done - Task 2: [name] — Done - Task 3: [name] — In progress, [what's done so far] </completed_work> <remaining_work> - Task 3: [what's left to complete] - Task 4: Not started - Task 5: Not started </remaining_work> <decisions_made> - Decided to use [X] because [reason] - Chose [approach] over [alternative] because [reason] </decisions_made> <blockers> - [Any blocker]: [status or workaround] </blockers> <context> [The mental model: what approach is being taken, why, what to watch out for] </context> <next_action> Start with: [the specific first action to take when resuming — be precise] </next_action> ``` Be specific enough that a fresh Cascade session can pick up immediately without re-asking. ## Step 4: Commit ```bash git add ".planning/phases/[phase_dir]/.continue-here.md" git commit -m "wip: [phase-name] paused at task [X]/[Y]" ``` ## Step 5: Confirm ``` ✓ Handoff created: .planning/phases/[phase_dir]/.continue-here.md Current state: - Phase: [phase_dir] - Task: [X] of [Y] - Status: in_progress - Committed as WIP ▶ To resume: resume-work ▶ Capture an idea before you go: /note [text] ▶ Generate a session summary: /session-report ``` --- ## Learning Checkpoint Read `learning_mode` from `.planning/config.json`. **If `auto`:** Offer before the session ends: > 💡 **Before you go:** Session transitions are when learning decays fastest. Two minutes now saves re-learning later: > > `@agentic-learning space` — Identify what was covered this session and schedule it for spaced review. Writes to `docs/revisit.md`. When you resume, you'll start sharper. > > `@agentic-learning reflect` — Quick 3-part reflection: what was built this session, what was the intent, what still feels uncertain. Takes 3 minutes, pays off when you return. **If `manual`:** Add quietly: *"Tip: `@agentic-learning space` before closing to schedule what you worked on for review."*

{ "prompt": "# Pause Work\n\nCreate a `.continue-here.md` handoff file that captures complete work state. Enables seamless resumption with full context restoration.\n\n**Use when:** stopping mid-plan, mid-phase, or any time you want to preserve exact position.\n\n## Step 1: Detect Current Phase\n\nFind the most recently active phase:\n```bash\nnode -e \"const fs=require('fs'),path=require('path');function find(d){let r=[];try{for(const e of fs.readdirSync(d,{withFileTypes:true})){const f=path.join(d,e.name);r=r.concat(e.isDirectory()?find(f):e.name.endsWith('-PLAN.md')?[f]:[]);}}catch(e){}return r;}const files=find('.planning/phases').map(f=>({f,t:fs.statSync(f).mtimeMs})).sort((a,b)=>b.t-a.t);if(files[0])console.log(files[0].f);\"\n```\n\nExtract the phase directory name from the result.\n\nIf no active phase detected, ask: \"Which phase are you pausing work on?\"\n\n## Step 2: Gather State\n\nCollect the complete handoff state by reading current files:\n\n```bash\ncat \".planning/phases/[phase_dir]/\"*-PLAN.md 2>/dev/null\ncat .planning/STATE.md\n```\n\nAlso ask the user (conversationally) for anything that can't be inferred from files:\n- Which specific task were you on?\n- What's already completed vs. what's in progress?\n- Any key decisions made this session that aren't yet committed?\n- Any blockers or things to watch out for?\n- What's the very first thing to do when resuming?\n\n## Step 3: Write Handoff File\n\nWrite `.planning/phases/[phase_dir]/.continue-here.md`:\n\n```markdown\n---\nphase: [phase_dir]\ntask: [current task number]\ntotal_tasks: [total tasks in current plan]\nstatus: in_progress\nlast_updated: [ISO timestamp]\n---\n\n<current_state>\n[Where exactly are we? Phase X, Plan Y, Task Z — immediate context for a fresh session]\n</current_state>\n\n<completed_work>\n- Task 1: [name] — Done\n- Task 2: [name] — Done\n- Task 3: [name] — In progress, [what's done so far]\n</completed_work>\n\n<remaining_work>\n- Task 3: [what's left to complete]\n- Task 4: Not started\n- Task 5: Not started\n</remaining_work>\n\n<decisions_made>\n- Decided to use [X] because [reason]\n- Chose [approach] over [alternative] because [reason]\n</decisions_made>\n\n<blockers>\n- [Any blocker]: [status or workaround]\n</blockers>\n\n<context>\n[The mental model: what approach is being taken, why, what to watch out for]\n</context>\n\n<next_action>\nStart with: [the specific first action to take when resuming — be precise]\n</next_action>\n```\n\nBe specific enough that a fresh Cascade session can pick up immediately without re-asking.\n\n## Step 4: Commit\n\n```bash\ngit add \".planning/phases/[phase_dir]/.continue-here.md\"\ngit commit -m \"wip: [phase-name] paused at task [X]/[Y]\"\n```\n\n## Step 5: Confirm\n\n```\n✓ Handoff created: .planning/phases/[phase_dir]/.continue-here.md\n\nCurrent state:\n- Phase: [phase_dir]\n- Task: [X] of [Y]\n- Status: in_progress\n- Committed as WIP\n\n▶ To resume: resume-work\n▶ Capture an idea before you go: /note [text]\n▶ Generate a session summary: /session-report\n```\n\n---\n\n## Learning Checkpoint\n\nRead `learning_mode` from `.planning/config.json`.\n\n**If `auto`:** Offer before the session ends:\n\n> 💡 **Before you go:** Session transitions are when learning decays fastest. Two minutes now saves re-learning later:\n>\n> `@agentic-learning space` — Identify what was covered this session and schedule it for spaced review. Writes to `docs/revisit.md`. When you resume, you'll start sharper.\n>\n> `@agentic-learning reflect` — Quick 3-part reflection: what was built this session, what was the intent, what still feels uncertain. Takes 3 minutes, pays off when you return.\n\n**If `manual`:** Add quietly: *\"Tip: `@agentic-learning space` before closing to schedule what you worked on for review.\"*", "variables": [], "notes": "Save a handoff file when stopping mid-phase so you can resume seamlessly later", "metadata": { "source": { "provider": "github-windsurf-workflows", "repository": "https://github.com/PauloHYBEX/Nexus-TCMS", "path": ".windsurf/workflows/pause-work.md", "ref": "af66d4e66d7443d045e74bafc302cd998ceb5b99", "url": "https://github.com/PauloHYBEX/Nexus-TCMS/blob/af66d4e66d7443d045e74bafc302cd998ceb5b99/.windsurf/workflows/pause-work.md", "key": "PauloHYBEX/Nexus-TCMS/.windsurf/workflows/pause-work.md" } } }

pause-work

About

Pause Work

Step 1: Detect Current Phase

Step 2: Gather State

Step 3: Write Handoff File

Step 4: Commit

Step 5: Confirm

Learning Checkpoint

Use it

Manifest

Reviews