planning-with-files
Plan Complex Tasks with Persistent Files
Également disponible depuis: ZhanlinCui,AmmarCodes,OthmanAdi
Long AI sessions cause context loss and forgotten goals. This skill uses markdown files as external memory to maintain task state across unlimited tool calls.
Télécharger le ZIP du skill
Importer dans Claude
Allez dans Paramètres → Capacités → Skills → Importer un skill
Activez et commencez à utiliser
Tester
Utilisation de "planning-with-files". Initialize planning for a Python CLI todo app with add, list, and delete commands
Résultat attendu:
- Created task_plan.md with 5 phases: Requirements, Planning, Implementation, Testing, Delivery
- Created findings.md with sections for requirements, research, and technical decisions
- Created progress.md with session timestamp and initial status
Utilisation de "planning-with-files". Log error: FileNotFoundError when reading config.json
Résultat attendu:
- Added to Errors Encountered table: FileNotFoundError, Attempt 1, Resolution: Create default config file
- Updated progress.md error log with timestamp
- Phase status remains in_progress until error resolved
Audit de sécurité
SûrAll static analysis findings were evaluated and determined to be false positives. The shell command patterns detected are from markdown documentation code blocks and legitimate bash helper scripts that only perform file operations. No cryptographic code, network exfiltration, or credential access exists. This is a safe productivity skill for planning complex tasks using persistent markdown files.
Facteurs de risque
⚙️ Commandes externes (2)
Score de qualité
Ce que vous pouvez construire
Multi-Step Development Projects
Track complex coding tasks across 50+ tool calls with phase tracking, error logging, and decision documentation.
Research and Documentation
Capture web search findings, browser results, and source materials in structured markdown before context is lost.
System Exploration and Debugging
Document discoveries, failed attempts, and resolutions when exploring unfamiliar codebases or debugging issues.
Essayez ces prompts
Create task_plan.md, findings.md, and progress.md files in my project directory. The goal is: [describe your task goal]. Start with Phase 1 for requirements gathering.
Read task_plan.md and update the current phase status. Log the error encountered: [paste error]. Document the root cause and attempted fixes in the Errors Encountered table.
I just completed two web searches on [topic]. Update findings.md with the key discoveries, relevant URLs, and any technical decisions made based on this research.
Mark Phase [X] as complete in task_plan.md. Summarize what was accomplished, files created or modified, and update progress.md with test results. Then set Phase [X+1] to in_progress.
Bonnes pratiques
- Create all three planning files before starting any complex task with 3+ steps
- Read task_plan.md before making major decisions to keep goals in attention
- Update findings.md immediately after every 2 view or browser operations
- Log every error encountered, even if quickly resolved, to prevent repetition
- Mark phase status changes explicitly: pending to in_progress to complete
Éviter
- Starting execution without creating a task_plan.md file first
- Silently retrying failed actions without logging the error
- Repeating the same failing action instead of mutating approach
- Creating planning files in the skill directory instead of project directory
- Using TodoWrite for persistence instead of file-based storage