Skills DirectorySkills Directory
SkillsLearnSecurityCategoriesDocsBlogPro
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
  • Chrome Extension
  • Skill Manager

Company

  • About
  • Community
  • Blog
  • API Docs
  • Advertise

2026 Skills Directory. All rights reserved.

ProTermsPrivacyRefunds
Back to skills

References

ASecurity

Everything needed to generate or edit an OpenCode skill. The host SKILL.md carries the METHOD (detect → route → load); this file carries the VALUES.

6 stars
0 votes
0 copies
0 views
Added 9/28/2026
ai-agentspythongoshellbashnodedockergitapi

Works with

claude codeapi

Security Analysis

A100/100

Scanned 9/28/2026

$npx -y skills add darellchua2/opencode-config-template --skill references --agent claude-code

Installs into .claude/skills of the current project.

Are you the author of References?

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

Security grade badge for References
[![Security: A — Skills Directory](https://www.skillsdirectory.com/api/skills/darellchua2-references/badge)](https://www.skillsdirectory.com/skills/darellchua2-references)

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

Files
skill.md
# Skill variant (values)

Everything needed to generate or edit an OpenCode skill. The host SKILL.md
carries the METHOD (detect → route → load); this file carries the VALUES.

## What this variant does

Create new OpenCode skills: frontmatter contract, lean content standard, directory placement, registry rebuild.

## The lean standard (house rule — supersedes "more detail is better")

A skill encodes **only what is house-specific**: triggers, conventions, version-pinned facts, workflow contracts, codified learnings, commands and paths that differ from upstream/vendor defaults. **Never re-teach what the model already knows** — no textbook explanations, no spec re-quotation, no vendor docs, no illustrative example catalogs.

Line budgets by category:

| Category | SKILL.md target | Guidance |
|---|---|---|
| Textbook/concept | 40–80 | learnings + triggers only; textbook deleted |
| Vendor API reference | 100–200 | pins, breaking changes, house conventions; recipes/tutorials deleted |
| House workflow | 40–60% of current | workflow steps + return contracts stay; ceremony/templates/examples go |
| Office/media | ≤200 + `reference.md` sibling | dense format reference moves to the sibling with a one-line pointer |

When in doubt, cut; a missing rule can be re-added from git history, but a bloated skill taxes every session that loads it.

## Frontmatter contract (runtime-read keys)

```yaml
---
name: <skill-name>          # required; MUST equal directory name; ^[a-z0-9]+(-[a-z0-9]+)*$, ≤64 chars
description: <1-1024 chars> # required; house style ≤50 words, preserve trigger phrases
license: Apache-2.0         # house default
compatibility: opencode
metadata:                   # optional string map; house sub-keys: protocol, pattern, os, harness (portability — see ## Portability)
category: <registry-group>  # installer-registry-only (build-registry/init/setup counts)
---
```

Unknown frontmatter fields are ignored by OpenCode. After ANY frontmatter change: `node installer/build-registry.mjs` and commit `registry.json`.

## Content structure

Minimum: `## What I do` (3–7 capability bullets), `## When to use me` (specific scenarios + not-for boundaries). Everything else only if it carries house signal. Skills live at root `skills/<name>/SKILL.md` (source of truth — never deployed copies).

**Body sections must be one of**: decision rules, boundaries (use-for / not-for), commands, templates, edge cases, workflow contracts. Worked examples: max 1 per genuinely non-obvious concept, none for stdlib-level knowledge — a modern model knows GoF patterns, pytest syntax, and Dockerfile idioms; it does not know your house rules. Descriptions carry capability + decision boundary ("not for X") + 3–6 trigger phrases; no API-name keyword stuffing.

## Skill permissions (v2)

Gate skills in the `permissions` array (config.json) or agent frontmatter — NOT inside SKILL.md:

```json
{ "permissions": [
  { "action": "skill", "resource": "*", "effect": "deny" },
  { "action": "skill", "resource": "<prefix>-*", "effect": "allow" } ] }
```

Last matching rule wins; deny-all first, allows after. The same `permissions` rules shape is used in config.json and agent frontmatter. `tools: skill: false` and the `permission.skill` map are deprecated.

## Portability

Skills ship to other harnesses (`--target claude|agents|kimi|kilo`) and OSes. Three house rules (full contract: repo `AGENTS.md` §Portability contract):

1. **Capability-binding block** — any harness-specific runtime mechanism (background shell, interactive question, subagent delegation) is written as a capability with per-harness rows plus a portable fallback:

   ```markdown
   <capability sentence>.
   - OpenCode: <mechanism>
   - Claude Code: <mechanism>
   - Other/none: <portable fallback — nohup+log-poll / plain-reply question / inline>
   ```

2. **Metadata vocabulary** (installer-only, zero runtime effect): double-quoted comma-separated strings, never brackets — `metadata.os: "linux, macos"` when the skill does not run everywhere; `metadata.harness: "opencode"` when the skill is about OpenCode itself.
3. **Bash rule** — every bash snippet carries `Requires bash (git-bash/WSL on Windows)` or becomes a `node -e` one-liner (Node is guaranteed: the installer is npx-based).

## File safety

ALWAYS `read` before `write`/`edit` on existing files — `write` overwrites silently. Use `edit` for targeted changes.

## Verification

```bash
python3 -c "import yaml; yaml.safe_load(open('skills/<name>/SKILL.md'))"  # YAML valid
grep '^name:\|^description:' skills/<name>/SKILL.md                        # required fields
node installer/build-registry.mjs --check                                   # registry in sync
```

> Removed 2026-09: the 7-step ceremony template, prompt templates, bash name-validation scripts, per-issue walkthroughs, and the padded example output — they were the regrowth engine. House contract verified against opencode.ai v2 docs in the repo's AGENTS.md §Skill / Agent Frontmatter Contract.

Attribution

darellchua2darellchua2
View sourceSee grades on GitHubMore from darellchua2 →
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

Terse caveman voice: answer first, fluff gone, every technical fact kept. Use for /caveman, "caveman mode", "talk like caveman", "be brief", "less tokens". Stays on until "stop caveman" or "normal mode".

1100021 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', ...

698621 votes

Writing Skills

Create and manage Claude Code skills in HASH repository following Anthropic best practices. Use when creating new skills, modifying skill-rules.json, understanding trigger patterns, working with hooks, debugging skill activation, or implementing progressive disclosure. Covers skill structure, YAML frontmatter, trigger types (keywords, intent patterns), UserPromptSubmit hook, and the 500-line rule. Includes validation and debugging with SKILL_DEBUG. Examples include rust-error-stack, cargo-dep...

3931 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.

3421 votes

catchup

Recovers the conversation and failed tool calls of a previous Codex, Amp, Claude Code, Antigravity, Cline, Copilot CLI, Cursor, DeepSeek Harness, Grok Build, 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.

741 votes
View all in ai-agents →