# Check Lower Compat Bounds > Test whether the lower compat bounds in Project.toml are still resolvable. Identifies stale bounds, dead version ranges, and suggests updates. Use when this capability is needed. - Skill: `tomevault-io/check-lower-compat-bounds` (Agent Skill, multi-file: 2 files) - Install (CLI): `npx skillmds@latest add tomevault-io/check-lower-compat-bounds` - Raw SKILL.md: https://api.skillmd.com/api/skills/tomevault-io/check-lower-compat-bounds/raw - Safety review: pending (external: skill-scanner PASS, skillspector PASS) - Works with: Claude Code, Claude.ai, OpenAI Codex - Category: Coding & Dev Tools - Author: tomevault-io (https://skillmd.com/u/tomevault-io) - Updated: 2026-09-17 - Page: https://skillmd.com/skills/tomevault-io/check-lower-compat-bounds --- You are checking the lower compat bounds of a Julia package for staleness. ## How it works The script `check_lower_compat_bounds.jl` (in the same directory as this skill) does the heavy lifting: - Parses all compat entries from Project.toml - For each dependency, tests whether the lower bound resolves in a fresh temp environment - Binary searches for the true minimum resolvable version when a bound fails - Reports dead ranges, stale bounds, and OK bounds - Saves suggestions to `.compat_suggestions.toml` and prompts to apply ## Steps 1. Run the script and let it complete (it takes a while, one temp environment per test): ``` julia .claude/skills/check-lower-compat-bounds/check_lower_compat_bounds.jl ``` 2. The script will print a report and prompt to apply changes. If the user declines, the suggestions are saved and can be applied later with: ``` julia .claude/skills/check-lower-compat-bounds/check_lower_compat_bounds.jl --apply ``` 3. Do NOT commit automatically. Let the user decide how to handle that. ## Important context - A bound that fails resolution is definitively stale: no user can install that version combination. This is caused by transitive dependency constraints, not bugs in this package. - A bound that passes resolution is not guaranteed to pass tests. If the user wants deeper validation, offer to run the test suite with specific deps pinned to their lower bounds. - The script skips stdlibs (LinearAlgebra, Printf, etc.) since those are tied to the Julia version. --- > Source: [nathanaelbosch/ProbNumDiffEq.jl](https://github.com/nathanaelbosch/ProbNumDiffEq.jl) — distributed by [TomeVault](https://tomevault.io).