Planning with Files
Persists agent working memory on disk via task_plan.md, findings.md and progress.md so long multi-step tasks survive context compaction and session breaks.
AutomationIntermediate★ 27,953⑂ 2,313AI score 10/10Last updated: Nov 7, 2026
What it does
- Before any complex task it initializes
task_plan.md(phases + status),findings.md(research), andprogress.md(session log) in your project directory — "working memory on disk." - Phases use a strict
### Phase N: Name+ single**Status:** pending/in_progress/completeformat, validated bycheck-complete.sh --validate. - Lifecycle hooks (UserPromptSubmit, PreToolUse, PostToolUse, Stop, PreCompact) re-inject the selected plan so state is recovered after compaction or a dropped session.
- Enforces practical rules: the 2-action rule (write findings after every two view/search ops), the 3-strike error protocol (diagnose → alternative → rethink → escalate), and an error log table.
set-active-plan.sh --listshows saved plans and phase progress; archiving happens only with explicit user approval.
Who it's for
- Anyone delegating work that needs 5+ tool calls: research, refactors, feature builds.
- Users whose agent drifts from the goal or repeats the same failed fix in long sessions.
- Teams running multi-session or parallel tasks that need separately pinned plans.
Example uses
- Multi-session feature work: initialize a "refactor payments module" plan with Phases 1–5; next day the hooks restore the current phase and findings automatically.
- Competitor research: after every two pages read, key numbers go into
findings.mdso multimodal/browser data is captured before it falls out of context. - Hard bug hunt: each attempt and error message is logged to the
## Errors Encounteredtable, preventing repeats and triggering a clear escalation after three failures.
· · · Install guide · · ·
Try it now, no install
Paste this into Claude to use the skill without installing anything.
Read the instructions in this file and follow them to help me: https://raw.githubusercontent.com/OthmanAdi/planning-with-files/HEAD/.cursor/skills/planning-with-files/SKILL.md What I want: (describe your task here)
If Claude can't open the link, open it yourself and paste the contents instead.
↓ If it works for you, download the ZIP below and install it. Then it runs on its own — no pasting each time.
Install in the Claude app (no terminal)
- Download the ZIP with the button below.
- In Claude, open Settings → Capabilities and turn on 'Code execution and file creation'. (one time)
- Go to Customize → Skills → + → 'Upload a skill' and upload the ZIP.
Install in Claude Code
Let Claude do it — paste this into Claude Code
Install the skill I found on AI Skill Mart. Copy the .cursor/skills/planning-with-files folder from the GitHub repo OthmanAdi/planning-with-files into my ~/.claude/skills/othmanadi-planning-with-files-4/. When it's done, tell me in one line what this skill can do.
Install with a command instead
git clone https://github.com/OthmanAdi/planning-with-files.git /tmp/pwf && mkdir -p ~/.claude/skills && cp -r /tmp/pwf/.cursor/skills/planning-with-files ~/.claude/skills/⚠ This is a third-party skill. Check the source repository before installing.
- Open a terminal and clone the repo:
git clone https://github.com/OthmanAdi/planning-with-files.git /tmp/pwf - Copy the skill into your user skills path:
mkdir -p ~/.claude/skills && cp -r /tmp/pwf/.cursor/skills/planning-with-files ~/.claude/skills/ - Make the helpers executable:
chmod +x ~/.claude/skills/planning-with-files/scripts/*.sh - Confirm Python exists (needed by the catch-up script):
python3 --version - If you installed to a non-standard path, add
export PWF_SCRIPT_DIR=~/.claude/skills/planning-with-files/scriptsto your shell profile so the hooks can find the script. - Restart Claude Code, then from your project root ask it to "create a plan for this task" — planning files appear under
.planning/. - On Windows, use the PowerShell variants:
init-session.ps1,check-complete.ps1,set-active-plan.ps1.
View source on GitHub ↗License: MIT