skills/pigeon/SKILL.md
Inter-session pmail - send and receive messages between Claude Code sessions running in different project directories. Uses global SQLite database at ~/.claude/pmail.db. Triggers on: mail, pmail, send message, check mail, inbox, inter-session, message another session, pigeon.
npx skillsauth add 0xDarkMatter/claude-mods pigeonInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
Inter-session messaging for Claude Code. Send and receive pmail between sessions running in different projects.
All commands go through MAIL, a shorthand for bash "$HOME/.claude/pigeon/mail-db.sh".
Set this at the top of execution:
MAIL="$HOME/.claude/pigeon/mail-db.sh"
Then use it for all commands below.
Parse the user's input after pigeon (or /pigeon) and run the matching command:
| User says | Run |
|-----------|-----|
| pigeon read | bash "$MAIL" read |
| pigeon read 42 | bash "$MAIL" read 42 |
| pigeon send <project> "<subject>" "<body>" | bash "$MAIL" send "<project>" "<subject>" "<body>" |
| pigeon send --urgent <project> "<subject>" "<body>" | bash "$MAIL" send --urgent "<project>" "<subject>" "<body>" |
| pigeon send --attach <path> <project> "<subject>" "<body>" | bash "$MAIL" send --attach "<path>" "<project>" "<subject>" "<body>" |
| pigeon reply <id> "<body>" | bash "$MAIL" reply <id> "<body>" |
| pigeon reply --attach <path> <id> "<body>" | bash "$MAIL" reply --attach "<path>" <id> "<body>" |
| pigeon broadcast "<subject>" "<body>" | bash "$MAIL" broadcast "<subject>" "<body>" |
| pigeon search <keyword> | bash "$MAIL" search "<keyword>" |
| pigeon status | bash "$MAIL" status |
| pigeon unread | bash "$MAIL" unread |
| pigeon list | bash "$MAIL" list |
| pigeon list 50 | bash "$MAIL" list 50 |
| pigeon projects | bash "$MAIL" projects |
| pigeon clear | bash "$MAIL" clear |
| pigeon clear 7 | bash "$MAIL" clear 7 |
| pigeon alias <old> <new> | bash "$MAIL" alias "<old>" "<new>" |
| pigeon purge | bash "$MAIL" purge |
| pigeon purge --all | bash "$MAIL" purge --all |
| pigeon id | bash "$MAIL" id |
| pigeon migrate | bash "$MAIL" migrate |
| pigeon init | bash "$MAIL" init |
When the user just says "check mail", "read mail", "inbox", "any mail?", or "any pmail?" - run bash "$MAIL" read.
When the user says "send mail to X", "send pmail to X", or "message X" - parse out the project name, subject, and body, then run bash "$MAIL" send.
Each project gets a stable 6-character hash ID derived from its git root commit (the very first commit in the repo). This means:
For non-git directories, falls back to a hash of the canonical path (pwd -P).
Use pigeon id to see your project's name and hash:
claude-mods 7663d6
When sending messages, you can address projects by name, hash, or path - they all resolve to the same hash ID.
Each project hash renders as a unique pixel-art identicon (11x11 symmetric grid using Unicode half-block characters). Run identicon.sh to see yours, or view all projects with pigeon projects.
A global PreToolUse hook checks for pmail on every tool call (no cooldown). Silent when inbox is empty.
=== PMAIL: 3 unread message(s) ===
From: some-api | Auth endpoints ready
From: frontend | Need updated types
... and 1 more
Use pigeon read to read messages.
Send file references with --attach <path> (repeatable). Paths are resolved to absolute and stored as references - files are not copied.
# Send with one attachment
pigeon send --attach src/config.ts my-api "Config update" "Updated the auth config"
# Send with multiple attachments
pigeon send --attach src/schema.sql --attach docs/API.md my-api "Schema + docs" "See attached"
# Reply with attachment
pigeon reply --attach output/report.json 42 "Here's the analysis"
Recipients see attachment paths with file sizes and can read them directly with the Read tool. If a file has been moved or deleted since sending, it shows as (missing).
touch .claude/pigeon.disable # Disable hook notifications
rm .claude/pigeon.disable # Re-enable
Only the hook is disabled - you can still send messages from the project.
Pigeon requires two things: scripts (the mail engine) and a hook (passive notifications). Both install globally - one setup, every project gets pmail.
sqlite3 - ships with macOS, most Linux distros, and Git Bash on Windows. No install needed.mkdir -p ~/.claude/pigeon
cp skills/pigeon/scripts/mail-db.sh ~/.claude/pigeon/
cp hooks/check-mail.sh ~/.claude/pigeon/
chmod +x ~/.claude/pigeon/mail-db.sh ~/.claude/pigeon/check-mail.sh
This gives you the pmail commands. You can now send and read messages manually:
bash ~/.claude/pigeon/mail-db.sh init # Create database
bash ~/.claude/pigeon/mail-db.sh status # Check it works
Add a hooks block to ~/.claude/settings.json. This makes Claude check for pmail automatically on every tool call:
{
"hooks": {
"PreToolUse": [
{
"matcher": "*",
"hooks": [
{
"type": "command",
"command": "bash \"$HOME/.claude/pigeon/check-mail.sh\"",
"timeout": 5
}
]
}
]
}
}
Important: If you already have a hooks section in your settings, merge the PreToolUse entry into the existing array - don't replace the whole block.
Without this step, pigeon still works but you have to check manually (pigeon read). With the hook, unread pmail appears automatically.
~/.claude/
settings.json # Hook config (you edit this)
pmail.db # Message store (auto-created on first use)
pigeon/
mail-db.sh # All pmail commands (send, read, reply, etc.)
check-mail.sh # PreToolUse hook (silent when inbox empty)
# Check your project identity
bash ~/.claude/pigeon/mail-db.sh id
# Send yourself a test message (use your project name from above)
bash ~/.claude/pigeon/mail-db.sh send "my-project" "Test" "Hello from pigeon"
# Check it arrived
bash ~/.claude/pigeon/mail-db.sh read
# Clean up
bash ~/.claude/pigeon/mail-db.sh purge --all
rm -rf ~/.claude/pigeon ~/.claude/pmail.db
# Then remove the hooks.PreToolUse entry from ~/.claude/settings.json
Single SQLite file at ~/.claude/pmail.db. Auto-created on first init or send.
CREATE TABLE messages (
id INTEGER PRIMARY KEY AUTOINCREMENT,
from_project TEXT NOT NULL, -- 6-char hash ID
to_project TEXT NOT NULL, -- 6-char hash ID
subject TEXT DEFAULT '',
body TEXT NOT NULL,
timestamp TEXT DEFAULT (datetime('now')),
read INTEGER DEFAULT 0,
priority TEXT DEFAULT 'normal'
);
CREATE TABLE projects (
hash TEXT PRIMARY KEY, -- 6-char ID (git root commit or path hash)
name TEXT NOT NULL, -- Display name (basename of project dir)
path TEXT NOT NULL, -- Canonical path
registered TEXT DEFAULT (datetime('now'))
);
| Issue | Fix |
|-------|-----|
| sqlite3: not found | Ships with macOS, Linux, and Git Bash on Windows. Run sqlite3 --version to check. |
| Hook not firing | Ensure hooks block is in ~/.claude/settings.json (Step 2 above) |
| Hook fires but no notification | Working as intended - hook is silent when inbox is empty |
| Messages not arriving | Target must be a known name, hash, or path. Use pigeon projects to see registered projects |
| Upgraded from basename IDs | Run pigeon migrate to convert old messages to hash-based IDs |
| Changed display name | Use pigeon alias old-name new-name to update the project's display name |
| Want to disable for one project | touch .claude/pigeon.disable in that project's root |
| Check your project ID | Run pigeon id to see name and 6-char hash |
tools
Behavioural-first software supply chain defense - catches poisoned npm/PyPI packages in the publish-to-advisory window that CVE tools miss. Use BEFORE every install or version bump (not only when an attack is suspected) - the 7-day cooldown gate + behavioural score catches freshly-published malware that CVE tools won't see for days. Socket.dev integration (free CLI + GitHub app + depscore MCP for Claude Code), stale-OIDC audit, dependency cooldown policy, publish-token rotation, VS Code extension audit, and a self-integrity scan that detects worm persistence hooks injected into Claude Code / VS Code settings. Triggers on: pip install, uv add, uv tool install, npm install, pnpm add, yarn add, cargo add, go get, composer require, gem install, upgrade dependency, dependency upgrade, version bump, bump version, bump package, adding dependency, new dependency, vetting a dependency, vet package, is this package safe, safe to install, should I install, before installing, pre-install check, preinstall scan, preinstall-check, PyPI cooldown, npm cooldown, release cooldown, minimumReleaseAge, score a package, package score, depscore, socket score, supply chain, supply chain attack, malicious package, poisoned dependency, npm worm, Shai-Hulud, behavioural scanning, Socket.dev, socket scan, dependency security, postinstall malware, OIDC token theft, compromised maintainer, typosquat, dependency confusion, package provenance, SLSA, persistence hook, malicious VS Code extension.
testing
GitHub remote operations — repo creation, metadata (description/homepage/topics), releases, README 'Recent Updates' enforcement, and issue / PR management with preview-before-send discipline. Companion to git-ops (local) and push-gate (pre-push safety). Three modes: new (first publish), update (subsequent release), audit (read-only checklist), plus atomic operations for issues and PRs. Triggers on: push to github, publish repo, ship release, cut release, gh release, set topics, repo description, github metadata, recent updates section, audit github repo, repo visibility, make repo public, gh repo create, gh issue, gh pr, create issue, comment on issue, close issue, triage issue, create PR, review PR, merge PR, pre-merge check, pr checks.
tools
Defend the agent's instruction surface against adversarial content - hidden-Unicode prompt injection (Trojan Source bidi reordering, U+E0000 tag-block ASCII smuggling, zero-width text), homoglyph confusables, and poisoned context that a human reviewer can't see but the model obeys. Scan CLAUDE.md / AGENTS.md / SKILL.md / .cursorrules and MCP tool descriptions; sanitize fetched web pages, issue/PR bodies, and dependency READMEs before they enter context. Triggers on: prompt injection, hidden unicode, invisible characters, zero-width space, bidi override, Trojan Source, ASCII smuggling, tag characters, homoglyph, confusable, unicode steganography, poisoned CLAUDE.md, malicious tool description, MCP tool poisoning, instruction injection, jailbreak in file, is this file safe, sanitize untrusted content, scan for hidden text.
tools
Set tool permissions for Claude Code. Configures allowed commands, rules, and preferences in .claude/ directory. Triggers on: setperms, init tools, configure permissions, setup project, set permissions, init claude.