Centralized management of AI coding agent skills, rules, and shell configs across multiple servers and providers.
canonical/ ← Single source of truth (edit here)
skills/ ← Workflow skills (plan, implement, review, evaluate, research)
rules/ ← Always-active rules (no-stub-mock, code-discipline, agent-behavior)
references/ ← Supporting docs referenced by skills
providers/ ← Auto-generated by transpile.py (DO NOT edit directly)
codex/ → ~/.codex/skills/
kiro/ → ~/.kiro/steering/ + ~/.kiro/skills/
claude-code/ → ~/CLAUDE.md
copilot/ → ~/.github/copilot-instructions.md
shell/ ← Shell configs (synced across servers)
zshrc → ~/.zshrc
bashrc → ~/.bashrc
agent-commands.sh ← CLI functions (sourced by both shells)
# Clone and install
git clone git@github.com:Himlamtech/AgentSkills.git ~/.agent-skills
cd ~/.agent-skills
python3 transpile.py
bash install.sh --full
source ~/.zshrcOr if agent-commands.sh is already sourced:
agent install# Edit a skill or rule
vim ~/.agent-skills/canonical/skills/plan.md
# Push changes to all servers
agent change "updated plan skill with new section"
# On another server, pull latest
agent update| Command | Description |
|---|---|
agent change [msg] |
Transpile + commit + push to GitHub |
agent update |
Pull + transpile + refresh symlinks |
agent install [url] |
First-time clone + full setup |
agent status |
Show git status |
agent diff |
Show uncommitted changes |
agent edit |
Open canonical/ in $EDITOR |
agent list |
List all skills, rules, references |
| Canonical | Codex | Kiro | Claude Code | Copilot |
|---|---|---|---|---|
rules/*.md |
Embedded in SKILL.md guardrails | steering/*.md (auto) |
Prepended to CLAUDE.md | Prepended to instructions |
skills/*.md |
<name>/SKILL.md |
skills/*.md (manual) |
Full body in CLAUDE.md | Summary + guardrails only |
references/*.md |
<skill>/references/ |
Referenced via #[[file:]] |
Skipped (too large) | Skipped |
- Create
canonical/skills/my-skill.mdwith frontmatter:--- name: my-skill description: > What this skill does and when to use it. triggers: - "trigger phrase 1" - "trigger phrase 2" ---
- Write the skill body in markdown
- Run
agent change "add my-skill" - On other servers:
agent update
- Create
canonical/rules/my-rule.mdwith frontmatter:--- name: my-rule description: > What this rule enforces. inclusion: always priority: high ---
- Write the rule body
- Run
agent change "add my-rule"