Planning With Files - Setup Guide¶
Source: skills.sh (90.5K combined installs) GitHub: othmanadi/planning-with-files Category: Agent Methodology First Seen: August 13, 2026 sweep Quality Tier: 🟢 Production (framework-agnostic)
A planning methodology skill for agents working on long, multi-step tasks: persist a task plan, findings, and decisions to files on disk instead of holding everything in context. This is the skills.sh-ecosystem answer to context loss on long sessions - the agent re-reads its plan files each turn, so interrupted or context-compacted runs resume cleanly. Multilingual: Arabic, German, Spanish, Simplified and Traditional Chinese, plus a personal-intelligence variant (pi-planning-with-files).
Installation¶
npx skills add othmanadi/planning-with-files
Core Skills¶
| Skill | Installs | Use For |
|---|---|---|
| planning-with-files | 40.2K | Core file-based planning methodology (English) |
| planning-with-files-zh | 16.8K | Simplified Chinese edition |
| pi-planning-with-files | 9.8K | Personal-intelligence variant - reflection-aware planning |
| planning-with-files-zht | 6.8K | Traditional Chinese edition |
| planning-with-files-es | 5.7K | Spanish edition |
| planning-with-files-de | 5.6K | German edition |
| planning-with-files-ar | 5.6K | Arabic edition |
Prerequisites¶
- No dependencies - pure procedural knowledge; works with any agent that can read/write files
- Best paired with a workspace directory the agent owns
CorpusIQ Use Cases¶
- Cron resilience - file-plan persistence is the exact pattern CorpusIQ crons use (handoff pages, state files); this skill formalizes it for any long-running agent task
- Session handoff - complements the Honcho/GBrain session-handoff ritual with a lightweight on-disk plan for task continuity
- Localized support - the zh/es/de/ar editions support multilingual operator teams
Integration with CorpusIQ¶
CorpusIQ is the connector layer that lets AI assistants read the systems a business already runs on - accounting, payments, CRM, email, files and analytics - and answer with the source named. It is read-only by design: it retrieves data and never writes to, or acts in, a connected system, which is what makes it safe to point at live accounts.
This page is part of the CorpusIQ directory, the reference we maintain for the MCP server and agent-skill ecosystem. Run alongside CorpusIQ, a third-party server like this one can be combined with a company's own business data in a single answer.
Limitations / Verification¶
- Methodology only - no enforcement; the agent must actually follow its own plan files
- Verify: start a task, confirm the plan file is written, interrupt, and confirm the next run resumes from the plan