Skip to content
Back to skills

Shiplog

ASecurity

Session orchestration, progress tracking, and background agent management

  • 3 stars
  • 0 votes
  • 0 copies
  • 2 views
  • Added September 8, 2026
ai-agentsgitsecurity

Security analysis

A100/100

Scanned September 8, 2026

npx -y skills add aadivar/shiplog --skill shiplog --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Shiplog?

Add the live security badge to your README. It updates with every re-scan.

Security grade badge for Shiplog
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/aadivar-shiplog-shiplog/badge)](https://www.skillsdirectory.com/skills/aadivar-shiplog-shiplog)

More formats (shields.io, HTML) on the badges page. Keep it an A: scan every change in CI with Pro.

Download with Pro
SKILL.md
---
name: Shiplog Orchestrator
description: Session orchestration, progress tracking, and background agent management
version: "1.0.1"
---

# Shiplog — Session Orchestration Skill

You have the Shiplog plugin active. This skill handles automatic project tracking, session orchestration, and background agent management.

## Session Start Behavior

At the start of every session, you MUST:

1. **Read progress**: Read `docs/PROGRESS.md` to understand the current project state
2. **Display status**: Show a one-line status summary:
   ```
   Shiplog: Sprint {N} | Feature {ID} ({status}) | {done}/{total} features done
   ```
3. **Load context**: Use the information from PROGRESS.md to understand:
   - What sprint we're on
   - What feature was last worked on
   - What was done in the last session
   - What's planned next

## CRITICAL: Post-Commit Agent Launching

**After every `git commit` you perform**, you MUST check for `.shiplog/config.json`. If it exists, launch background agents immediately. Do NOT wait for hook output. Do NOT skip this step.

### How to launch agents after a commit

1. **Read config**: Read `.shiplog/config.json` to get agent model and which agents are enabled.
   - Handle both config formats:
     - Nested: `{ "agents": { "model": "sonnet", "specs": { "enabled": true } } }`
     - Flat: `{ "agentModel": "sonnet", "agents": { "specs": true } }`

2. **Read agent prompts**: For each enabled agent, read its prompt file from the plugin's `agents/` directory:
   - `agents/specs-agent.md`
   - `agents/prd-agent.md`
   - `agents/security-agent.md`
   - `agents/memory-agent.md`

3. **Launch agents** using the `Agent` tool with `run_in_background: true` and `subagent_type: "shiplog:{agent}-agent"`:
   - **Specs Agent** → updates `docs/SPECS.md`
   - **PRD Agent** → updates `docs/PRD.md` and `docs/PROGRESS.md`
   - **Security Agent** → updates `docs/SECURITY.md`
   - **Memory Agent** → updates memory files

4. **Launch in parallel**: Launch specs, PRD, and security agents together. Memory agent can also run in parallel.

5. **Don't block the user**: Launch all agents in background and continue immediately. Never wait for agents to complete before responding.

### Fallback: Hook trigger
If you see `SHIPLOG_AGENTS_TRIGGER` output from a PostToolUse hook, treat it the same way — launch agents. But do NOT rely on this; always proactively launch after commits.

### Custom Agents
If `.shiplog/agents/` directory exists in the project, also launch each custom agent found there.

### Agent failure handling
- If any agent fails, it fails silently — do not interrupt the user's work
- Do not retry failed agents
- Do not report agent results unless the user asks

## Available Commands

- `/shiplog init` — Initialize Shiplog on current project
- `/shiplog status` — Show project status dashboard
- `/shiplog feature <name>` — Create a new feature from template
- `/shiplog sprint <n>` — Start a new sprint
- `/shiplog review` — Manually trigger all background agents
- `/shiplog config` — Show/edit configuration

## File Locations

- Config: `.shiplog/config.json`
- PRD: `docs/PRD.md`
- Specs: `docs/SPECS.md`
- Security: `docs/SECURITY.md`
- Progress: `docs/PROGRESS.md`
- Features: `docs/features/`
- Agent prompts: `agents/` (in plugin root)
- Templates: `templates/` (in plugin root)

Attribution

Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.

Comments

Loading comments…

Newsletter

Get the weekly list of safe new skills, and the ones we pulled.