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

Superstack Smith

ASecurity

A new skill or pack for a superstack-family plugin — "new skill", "add a module", "build a pack for X" — or a recurring moment applying for admission. Skip for editing an existing skill (just edit, red-first near hooks) and product ideas that aren't skills (superstack-inception).

2 stars
0 votes
0 copies
1 views
Added 9/19/2026
ai-agentsgoshell

Security Analysis

A100/100

Scanned 9/19/2026

Install to Claude Code

$npx -y skills add debabsah/superstack --skill superstack-smith --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of Superstack Smith?

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

Security grade badge for Superstack Smith
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/debabsah-superstack-smith/badge)](https://www.skillsdirectory.com/skills/debabsah-superstack-smith)

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

Download with Pro
Files
SKILL.md
---
name: superstack-smith
description: A new skill or pack for a superstack-family plugin — "new skill", "add a module", "build a pack for X" — or a recurring moment applying for admission. Skip for editing an existing skill (just edit, red-first near hooks) and product ideas that aren't skills (superstack-inception).
---

# superstack-smith

The smith is a **gatekeeper, not a generator** — it exists to make a bad skill hard to add. The standing law is "no module without its trigger"; the smith is that law's door, mechanized. An idea that fails a gate parks in the queue with `revisit: <the evidence wanted>` — parking is a pass-through outcome, not a rejection.

## The admission gates, in order

1. **The evidence line — first, non-negotiable.** What recurring *moment*, observed *where* (sessions, logs, the user's own words, a hand-rolled workaround on disk). A hunch parks; observed recurrence proceeds. The strongest evidence is a workaround the user already built by hand — it proves demand *and* sketches the shape.
2. **The moment names its action class** — build / design / publish / resume / correction / broken-live / choosing / reporting-out — never a session size, never a topic. The taxonomy check runs here: **topics get review lenses or oracle rows; rules get lines in an existing skill; only moments get runners.** Check both cheaper homes before minting a skill.
3. **Collision check, mechanically.** Run it, don't recall it:
   ```
   find skills ~/.claude/skills ~/.claude/plugins -name SKILL.md \
     -exec grep -h '^description:' {} + 2>/dev/null
   ```
   (`find` tolerates roots that don't exist on this machine — a shell glob does not: an unmatched glob aborts the whole command under zsh and returns a silent, empty "no collisions". Adjust roots as needed; follow symlinks to the real trees.) Then hunt the proposed trigger's verbs and nouns in the output. A real overlap means widen-or-point-at the existing skill instead. The nearest *sibling* goes in the new description's Skip clause **by name**; a third-party neighbor goes in as a **class with an example** — "a spec tool (e.g. another pack's `spec` skill) when installed" — never a bare dependency, so the description degrades gracefully on machines without the pack. Routing ambiguity is a dense skill environment's #1 scaling risk, and the Skip clause is its cheapest control.
4. **State check.** If it owns durable state, name its file under `.superstack/` (one ruled exception on record: superstack-decide's committed `docs/decisions/` — the README names it and says why) and its lifecycle — append-only? expiry? who compacts? **and who reads it** (a write with no reader is ritual) — and **reuse an existing grammar** (queue `Q<n>`, gates `G<n>`, value `V<n>`, task files, ledger tokens) before minting a new one. A bespoke new ledger format is a smell with a history here.
5. **Loop check.** Name where it hands off into the existing loop — scope, verify, ship, debug, queue, a premise ledger. A skill with no edge into the loop is an island, and probably a topic wearing a skill's clothes.
6. **Gate and injector check.** Needs session-start presence? Ride a task-file pointer (`<!-- task: <slug> — goal: … — next: … -->`) — the existing hook surfaces it *within its line budget*, so presence competes with other tasks; a `00-` filename prefix wins the sort when presence is critical. An actual hook edit is red-first with its suite, and kernel hooks change upstream only, never in place.

## Writing it (house shape)

- Frontmatter description: `Use when <the moment, concretely>… Also on "<name>". Skip for <nearest neighbor by name + the trivial case>.`
- Body ≤ ~40 lines: principle first, numbered steps, existing grammars **cited, not restated** — a runner echoes canon and adds only its delta (doctrine stated twice drifts apart).
- Packs live in their **own plugin**, never in this core — a daily driver carries no profession's specialty gear.

## Shipping it

`claude plugin validate .` green; the plugin's suites still green — **including `hooks/test-front-doors.sh`**, which mechanically asserts the new module's full name appears in every enumerated surface (`.claude-plugin/plugin.json`, `.claude-plugin/marketplace.json`, the README table). The front door rots exactly when this step is a memory, and it did once (0.5.0's marketplace shipped saying "nine modules" with eighteen present) — hence the check, not the reminder. Then a DECISIONS.md entry (chose / rejected / who / reopens-if) records the admission — a module that can't say what evidence admitted it can't say what evidence would retire it.

Attribution

debabsahdebabsah
View sourceMore from debabsah →
SSkills DirectorySkills Directory

Your tool, in front of Claude Code builders.

3 founder slots · $299/mo · GSC-verified traffic · sponsors can never buy grades.

See placements

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

Your tool, in front of Claude Code builders.

3 founder slots · $299/mo · GSC-verified traffic · sponsors can never buy grades.

See placements

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".

1066601 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', ...

686011 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.

651 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 →