Plannotator Compound

If you've been using Plannotator to gate your Claude Code sessions, this skill reads your archive of denied plans and generates an HTML dashboard showing why you keep getting rejected. It processes every denied markdown file (or falls back to Claude Code's ExitPlanMode logs if you don't have Plannotator data), extracts denial patterns and feedback taxonomy, and produces actionable prompt improvements. The incremental mode is smart: it detects previous reports by version number and lets you analyze only plans since the last run, which saves tokens on large archives. Most interesting part is the batching strategy. It spawns parallel Haiku agents scaled to your dataset size, from 2 agents for small batches up to however many needed to keep each one processing 10-15 files.

skills.sh installs: 12.8K
GitHub stars: 10K
last updated: Aug 21, 2026

View on GitHub

Install

$ npx -y skills add backnotprop/plannotator --skill plannotator-compound --agent claude-code

Installs into .claude/skills of the current project.

Files

Compound Planning Analysis

You are conducting a comprehensive research analysis of a user's Plannotator plan archive. The goal: extract patterns from their denied plans, reduce them into actionable insights, and produce an elegant HTML dashboard report.

This is a multi-phase process. Each phase must complete fully before the next begins. Research integrity is paramount — every file must be read, no skipping.

Source Selection

Before starting the analysis, determine which data source is available.

Plannotator mode (first-class) — Find the Plannotator data directory, called <data-dir> from here on. Run:
```
plannotator data-dir
```
It prints the folder's absolute path and creates nothing; use what it prints. If it prints nothing or fails (an older Plannotator), use $PLANNOTATOR_DATA_DIR if set, else whichever of $XDG_DATA_HOME/plannotator and the home folder's .plannotator holds history/ or sessions/ (the home one if both do), and ask the user if neither does.