Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsCommunityBlog
Sign InSubmit Skill
Skills Directory

Security-tested agent skills for Claude, coding agents, and AI workflows.

Directory

  • Browse Skills
  • All Skills A–Z
  • Claude Skills
  • Claude Code Skills
  • Agent Skills
  • Categories
  • Authors
  • Submit a Skill

Learn

  • Learn Hub
  • Install Claude Skills
  • Write SKILL.md
  • Skills vs MCP
  • Directories Compared

Security

  • Security
  • Methodology
  • Secure Claude Skills
  • Security Badges

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

Seal

ASecurity

Ships a feature — simplify pass, certify, commit, PR, CI monitoring, review loop, merge. Use when a feature branch is verified and ready to ship.

14 stars
0 votes
0 copies
0 views
Added 9/28/2026
ai-agentsgobashgitapidocumentation

Works with

claude codeapi

Security Analysis

A100/100

Scanned 9/28/2026

Install to Claude Code

$npx -y skills add Alexander-Tyagunov/magician --skill seal --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Seal?

Add the live security badge to your README — it updates automatically with every re-scan.

Security grade badge for Seal
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/alexander-tyagunov-seal/badge)](https://www.skillsdirectory.com/skills/alexander-tyagunov-seal)

More formats (shields.io, HTML) on the badges page.

Files
SKILL.md
---
name: seal
description: Ships a feature — simplify pass, certify, commit, PR, CI monitoring, review loop, merge. Use when a feature branch is verified and ready to ship.
allowed-tools: Read, Task, AskUserQuestion, Edit(./CLAUDE.md), Edit(./README.md), Bash(git add -A), Bash(git commit -m *), Bash(git push -u origin HEAD), Bash(gh pr create *), Bash(gh pr checks *), Bash(gh pr view *), Bash(gh pr merge --squash --delete-branch), Bash(gh run view *)
argument-hint: "[pr-title]"
---

# /seal — Ship to Production

Take a certified feature branch through to a merged PR. This skill performs irreversible actions (push, PR, merge) — it presents a single consolidated ship-summary and waits before the first outward command (`git push`).

## Pre-flight

- Confirm /certify has been run and passed
- Confirm the /scrutinize review + remediation cycle is complete (or explicitly skipped by user)
- **Never ship on stale evidence** — /certify must have passed in *this* session (fresh output you read this turn), not a remembered run. See [lore/verification.md](../../lore/verification.md).

## Autonomy — approve the plan, then run

Once Pre-flight is confirmed (/certify passed, /scrutinize + remediation complete or explicitly skipped), the ship sequence runs **autonomously**: the Simplifier Pass, Final Certify, Update Documentation, and Commit steps proceed without stopping to ask the owner. Claude Code's built-in read-only commands (including read-only git) don't prompt; this skill pre-approves the exact commit, push, PR, CI-check and merge commands shown below and edits to `CLAUDE.md`/`README.md`; any other edit or command goes through the normal permission prompt unless the user runs in auto mode. Re-gate **only** on outward side effects — `git push`, `gh pr create`, `gh pr merge` — surfaced once through the consolidated ship-summary gate below, not per command. Doctrine: [lore/autonomy.md](../../lore/autonomy.md).

## Process

### 1. Simplifier Pass
Dispatch the simplifier via `Task` with subagent type `magician:simplifier` for a final simplification sweep. Give it a self-contained prompt: the goal, the changed files (with diff), and the return format (see [lore/subagent-context.md](../../lore/subagent-context.md)). Fix any Important findings. Skip Low suggestions unless trivial.

### 2. Final Certify
Run /certify. All checks must pass before continuing — including style vs the project's documented conventions ([lore/code-standards.md](../../lore/code-standards.md)), so review bounces on style don't reopen the ship loop.

### 3. Update Documentation
Before committing, update docs to reflect the shipped feature:
- **CLAUDE.md** — add or update any changed commands, env vars, architecture notes, or setup steps
- **README.md** — update feature list, screenshots, or usage examples if the change is user-visible
- Any other docs that reference the changed code (API docs, architecture diagrams, etc.)

Keep updates minimal and accurate — document what changed, not the implementation details.

### 4. Commit
```bash
git add -A
git commit -m "feat: <feature description>

<bullet list of what was built>

Co-Authored-By: Claude <noreply@anthropic.com>"
```

### Ship-summary gate — one outward approval
Steps 1–4 ran autonomously. Before the first outward command (`git push`), present **one** consolidated ship-summary, then gate the outward push/PR/merge through **AskUserQuestion** — end your turn at the call and act on the choice. The summary covers:
- **Changed files** — the diff going out (from step 4's `git add -A`)
- **Commit message** — what was committed in step 4
- **PR title + body** — the title plus the Summary + Test plan from step 6
- **Merge strategy** — e.g. `--squash --delete-branch` (step 9)

Ask **"Ship this branch?"** with these options:
- **Ship it** — run Push, Create PR, and Merge without asking again
- **Revise** — adjust the commit, PR, or scope first; re-present the summary afterward
- **Cancel** — stop here; do not push, open a PR, or merge

Treat a free-form "yes / approved / looks good" as **Ship it**. On **Ship it**, run Push (step 5), Create PR (step 6), and Merge (step 9) without asking the owner again.

### 5. Push
```bash
git push -u origin HEAD
```

### 6. Create PR
Use the PR title + body approved at the ship-summary gate above — no separate title prompt:
```bash
gh pr create --title "<title>" --body "$(cat <<'EOF'
## Summary
- <what was built>
- <key decisions>

## Test plan
- [ ] All tests pass
- [ ] Types clean
- [ ] Lint clean
- [ ] Manually verified: <golden path>

Built with magician + Claude Code
EOF
)"
```
The `$(…)` body can keep this command from matching the `gh pr create` pre-approval; if Claude Code asks to confirm it, that is its own permission check, separate from the ship-summary gate.

### 7. Watch CI (evaluator-optimizer loop)
Run the checks watcher with the Bash tool's background option, so the turn isn't held open; you're notified when it exits, then read its output:
```bash
gh pr checks <pr> --watch
```

Then loop: **on a failing check → `gh run view <id> --log-failed` → fix → commit → `git push -u origin HEAD` → start the watcher again → repeat until every check is green.** For a long or unattended wait, pair with **`/goal`** ("PR checks green, then merged") so Claude keeps driving across turns; on a schedule, `/loop check CI on my PR` (self-paces when you omit the interval).

### 8. Review Comments
If reviewers add comments: use /scrutinize to process and remediate them, then /certify, then `git push -u origin HEAD`.

### 9. Merge (if auto-merge not enabled)
```bash
gh pr merge --squash --delete-branch
```

## disableGit Mode

If `disableGit: true`: skip push/PR/merge steps. Commit locally and report done.

## Obstacles

If this skill runs as a dispatched unit (under /orchestrate, /weave, /manifest, /transmute, or another skill) and hits something that blocks or degrades the work, do not wait for a human who is not there and do not silently ship a degraded result — return an Obstacles block to the caller, alongside whatever you did complete:

```
STATUS: BLOCKED | DEGRADED | NEEDS_CONTEXT
OBSTACLE: <one-line label of what blocked or degraded the task — the claim alone>
BLOCKER: <the specific, actionable cause — distilled, never a raw traceback or dumped log>
SEVERITY: Critical | High | Medium | Low
WORKAROUND: <what you did to proceed and what it leaves unverified; empty if still fully blocked>
RECURRENCE: First-seen | Recurring | Systemic
SCOPE: <this task only | likely hits sibling/downstream work too>
NEXT: <the action or decision the caller must make to clear it — retry with X, supply input Y, accept degraded, or escalate>
```

When invoked interactively by a human, surface the same obstacle in prose instead. Omit the block entirely on a clean run. See [lore/obstacles.md](../../lore/obstacles.md).

## Completion Signal

"Sealed. PR merged. Run /portal cleanup steps if applicable."

Attribution

Alexander-TyagunovAlexander-Tyagunov
View sourceMore from Alexander-Tyagunov →
SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

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 (0)

No comments yet. Be the first to comment!

SSkills DirectorySkills Directory

Ship a skill? Prove it's safe.

Free 120-pattern security scan, letter grade, and an embeddable README badge.

Submit a skill

Related Skills

Caveman

Ultra-compressed communication mode that cuts output tokens while keeping technical accuracy. Levels: lite, full, ultra and the wenyan variants. Use for /caveman, "caveman mode", "talk like caveman", "be brief" or "less tokens".

1074701 votes

Hyperplan

Adversarial multi-agent planning skill. Self-orchestrates 5 hostile category members (unspecified-low, unspecified-high, deep, ultrabrain, artistry) via team-mode for ruthless cross-critique debate, distills only the defensible insights, then MANDATORILY hands the distilled insight bundle to the `plan` agent for executable plan formalization. Use when planning needs maximum rigor and surfacing of weak assumptions, blind spots, and over-engineering. Triggers: 'hyperplan', 'hpp', '/hyperplan', ...

695601 votes

Mcp Code Execution

Routes multi-tool workflows through MCP servers for large datasets and pipelines. Use when Bash tool overhead is limiting throughput on data-heavy tasks.

3351 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Kimi, OpenCode, Pi Agent, or ZCode session. Use when the user says "catch up", "what did the last session do", "get me up to speed", "I switched agents", asks to recover/summarize a previous session before continuing, or asks to diagnose or report a catchup failure. Do NOT use for the current conversation, git history, or any non-agent log.

691 votes

math-skill

A comprehensive mathematical reasoning skill for AI assistants — handles arithmetic to research-level problems with rigorous step-by-step reasoning, systematic verification, and transparent uncertainty handling

381 votes
View all in ai-agents →