SkillsAdd

smithery/grahama1970

skills-broadcast

Sync skills across all IDEs (Pi, Codex, Claude Code, KiloCode) using symlinks to a single canonical directory. Use "sync skills", "broadcast skills", or "link skills".

Installation

$ npx skills add smithery/grahama1970 --skill skills-broadcast
View on skills.sh ↗ Website ↗

Similar popular skills

Related neighbors and high-traction skills in the same topics — useful to compare before installing.

broadcast-campaign Similar name
zavudev/zavu-skills

Create and manage broadcast campaigns for bulk messaging across SMS, WhatsApp, Email, and Teleg…

211 installs
live-broadcast Similar name
mnhkahn/cyeam-cli

直播预报——NBA/世界杯/中国男女足比赛赛程与转播源。用户问比赛，直接跑 cyeam tv 命令返回结果，不要…

105 installs

Also in this package

Other skills from smithery/grahama1970 · top by installs.

npx skills add smithery/grahama1970
youtube-transcripts 0 security-scan 0 scillm 0 review-music 0 ops-sam-gov 0 fetcher 0 episodic-archiver 0 create-storyboard 0 create-movie 0 create-image 0

Browse all from smithery/grahama1970

More details

Agent compatibility

Declared targets from SKILL.md / docs. Unmarked agents are not listed — the skill may still install via the CLI.

Claude Code Declared
Cursor Not declared
Codex Declared
GitHub Copilot Not declared
Windsurf Not declared
Gemini CLI Not declared
Cline Not declared
OpenCode Not declared

Skill metadata

Parsed from SKILL.md frontmatter.

Allowed toolsBash, Read
Declared agents claude-code codex
More metadata
short-description
Symlink-based skill sharing across all IDEs

Package contents

Files included with this skill beyond the listing page.

  • skill md SKILL.md 6,072 B
  • docs SUMMARY.md 185 B

History

  1. First recorded snapshot · 0 installs

SKILL.md

STOP. READ THIS ENTIRE SKILL.MD BEFORE CALLING ANY ENDPOINT.

Skills Broadcast Skill

Share skills across all IDEs and projects using symlinks to a single canonical directory.

Architecture

┌─────────────────────────────────────────────────┐
│        CANONICAL (Source of Truth)               │
│   <project>/.pi/skills/                         │
│   (the one real copy — edit here)               │
└───────────────────────┬─────────────────────────┘
                        │
              symlinks (ln -sfn)
                        │
    ┌───────────┬───────┼───────┬───────────┐
    ▼           ▼       ▼       ▼           ▼
~/.codex/   ~/.claude/ ~/.pi/  ~/.kilocode/ project/
 skills      skills    agent/   skills      .agent/
                       skills               skills

Every target is a symlink. No copies, no rsync, no sync lag.

Why Symlinks?

The old rsync approach:

  • Duplicated ~300 GB across 10+ targets
  • Required 500+ lines of rsync exclusion lists
  • Caused the 2026-02-11 deletion incident (rsync --delete propagated a wiped dir)
  • Changes required manual push to propagate

Symlinks:

  • 0 bytes duplicated
  • ~200 lines of code
  • Changes are instant everywhere
  • Impossible to accidentally delete skills via sync

Quick Start

# See current state
./run.sh status

# Create symlinks at all targets (safe — backs up old dirs)
./run.sh link --dry-run    # preview first
./run.sh link              # do it

# Legacy commands still work
./run.sh push    # same as link
./run.sh pull    # same as link

Commands

Command Description
./run.sh link Create symlinks at all targets
./run.sh status Show all targets and their link state
./run.sh git-sync Commit and push skills to agent-skills GitHub repo
./run.sh cleanup Delete .pre-symlink-* backups to reclaim disk
./run.sh register PATH Add a project to target registry
./run.sh unregister PATH Remove a project from registry
./run.sh targets List registered projects
./run.sh push Legacy alias for link
./run.sh pull Legacy alias for link

Status Output

=== Skills Broadcast Status ===

Canonical: /home/user/workspace/pi-mono/.pi/skills (146 skills)

Project: /home/user
  OK      .pi/agent/skills -> canonical
  OK      .codex/skills -> canonical
  OK      .claude/skills -> canonical
Project: /home/user/workspace/experiments/memory
  OK      .pi/skills -> canonical

Safety

  • Old directories are backed up (renamed with .pre-symlink-<timestamp>) before replacement
  • Canonical must have 20+ skills or the script refuses to run
  • --dry-run previews all changes without modifying anything
  • Backups can be deleted after verification: rm -rf path.pre-symlink-*

Registering Projects

# Register a project for symlink targets
./run.sh register /path/to/project

# The registry is at ~/.agent_skills_targets
cat ~/.agent_skills_targets

Supported IDEs

IDE Skill Location Pattern
Pi ~/.pi/agent/skills .pi/skills
Codex ~/.codex/skills .codex/skills
Claude Code ~/.claude/skills .claude/skills
KiloCode ~/.kilocode/skills .kilocode/skills
Generic .agent/skills .agent/skills

Git Sync to GitHub

The git-sync command pushes canonical skills to the grahama1970/agent-skills repo on GitHub. This is a manual operation — run it when you want to checkpoint your current skill state.

./run.sh git-sync --dry-run            # preview inferred single-skill commit
./run.sh git-sync --skill ask          # commit and push one skill
./run.sh git-sync --skill ask --skill subagent-runner
./run.sh git-sync --all                # intentionally commit all changed skills

How it works:

  1. Refuses to run if unrelated changes are already staged.
  2. Infers a single changed skill, or requires explicit --skill NAME / --all.
  3. Stages only selected skill paths with runtime artifacts excluded.
  4. Commits with a scoped message and pushes normally; it does not force-push.

This command is intentionally conservative for parallel-agent worktrees. Do not use broad staging (git add -A) when multiple agents may have uncommitted work. It also supports normal Git worktrees where .git is a file rather than a directory.

Heavy Artifact Policy

Heavy artifacts (models, weights, checkpoints, datasets) must live on /mnt/storage12tb/ and be symlinked into skill dirs. The sanity check enforces this with a 100 MB threshold per subdirectory.