Nanoskills

NanoSkills

Use this skill as the package-level catalog and help system for Awesome NanoClaw Skills.

/nanoskills is discovery and documentation. It should be fast, friendly, offline-first, and written in the language of the current conversation.

The catalog is generated from this package's skills/*/SKILL.md frontmatter during repository validation. In a NanoClaw container, treat installed skill files as read-only and only list first-party Awesome NanoClaw Skills from the generated catalog files.

Managed Auto Update

If this skill is installed as a managed Awesome NanoClaw Skill and ../awesome-updater/scripts/awesome_skills.py exists, run this before the normal workflow:

python3 ../awesome-updater/scripts/awesome_skills.py check awesome-updater --skills-dir .. || true
python3 ../awesome-updater/scripts/awesome_skills.py discover --skills-dir .. || true
python3 ../awesome-updater/scripts/awesome_skills.py check nanoskills --skills-dir .. || true

Continue the normal workflow if the update or discover check fails because of network, GitHub, or local environment issues.

Do not run extra package-wide update checks during normal catalog or help responses beyond the shared managed preamble above. Listing skills should be based on local catalog files and should not require network access to produce the answer.

Triggers

Run this skill when the user invokes or asks for:

• /nanoskills
• /nanoskills list
• /nanoskills help
• /nanoskills help
• "what skills are available?"
• "help for the NanoClaw skills"

Load References

• Read references/catalog.md before listing skills.
• Use references/catalog.json only as resolver metadata for skill-specific help.
• Do not list aliases from catalog.json in normal catalog or normal help output. Aliases are private routing metadata.
• If the generated catalog files are missing, fail closed: say the package catalog is unavailable and recommend updating or reinstalling nanoskills.
• Never scan sibling runtime skill folders, global Claude/Codex skill folders, external package folders, memory, inferred capabilities, or chat history to build the catalog.
• Use templates/help-response.md when explaining one skill.

Language

Answer in the language of the current conversation. If the conversation is mixed, follow the user's latest message.

Do not translate slash commands, skill folder names, file paths, or literal command examples.

Command Router

Choose the response from the user's command:

| Input | Behavior | | --- | --- | | /nanoskills | Show the full package catalog. | | /nanoskills list | Show the full package catalog. | | /nanoskills help | Explain how to use /nanoskills itself and show how to request help for one skill. | | /nanoskills help | Show detailed help for the requested skill. | | /nanoskills updates or /nanoskills status | Explain the relationship with awesome-updater and show safe update/status commands. Use updater tooling only when the user clearly asks to discover, check, install, configure, or upgrade. |

Accept skill names, aliases, and slash commands when resolving help:

• Prefer canonical skill names and primary slash commands from the generated catalog.
• Use aliases from references/catalog.json only to resolve a user's requested help target.
• If an alias resolves to a skill, explain the canonical command for that skill and do not promote the alias as a public command.

If the target skill is ambiguous or unknown, show the catalog and ask the user to choose one by name or slash command.

Catalog Output

For /nanoskills, produce a scannable catalog:

1. one short sentence explaining that this is the Awesome NanoClaw Skills package;
2. User-facing skills section;
3. Admin and package skills section;
4. for each skill:

• slash command;
• what it does;
• when to use it;
• expected output;
• one compact example;

1. close with examples of help commands.

Keep the catalog clear and visually pleasant, but do not use tables when the current chat surface is likely Telegram or mobile chat.

The normal catalog output must include only primary slash commands from references/catalog.md. Do not add compatible aliases, legacy aliases, global commands, runtime commands, health-check commands, or capability summaries that are not present as primary commands in the generated catalog.

Help Mode

For /nanoskills help , use templates/help-response.md.

Detailed help must include:

• what the skill does;
• when to use it;
• when not to use it, if important;
• command forms, using primary slash commands only unless the user explicitly asks about an alias;
• expected input;
• expected output;
• curated examples;
• contextual examples when the current conversation has useful context.

Contextual Examples

Help examples are hybrid:

• Always include curated examples from references/catalog.md or the target skill's SKILL.md.
• Add a Contextual examples section only when the visible conversation gives enough concrete context to create useful examples.
• Contextual examples should adapt the user's current topic, objects, filenames, links, decisions, or tasks into valid slash-command examples.
• Do not quote secrets, tokens, private URLs, long transcripts, or sensitive conversation content.
• If context is weak, omit contextual examples instead of inventing generic personalization.

Examples:

• If the conversation is about launch timing, /febaboard Should we launch now or wait until onboarding is better?
• If the conversation is about a repo architecture decision, /rethink Is this global command architecture simple enough?
• If the conversation includes a URL, /readthis

Auto-Update Relationship

/nanoskills is not the updater. It should explain and route update-related intent without merging responsibilities.

• Use /nanoskills and /nanoskills help for discovery and usage help.
• Use awesome-updater for install, discover, check, config, backup, rollback, and auto-upgrade behavior.
• Normal catalog/help responses must not run extra package-wide checks beyond the shared managed preamble.
• If the user asks to check updates, use awesome-updater instructions or tooling when available.
• If the user asks for a pure explanation, do not run update commands.

Output Rules

• Match the language of the current conversation.
• Keep answers concise but complete enough that a new user can choose the right skill.
• Prefer commands and examples over abstract explanation.
• Do not claim a skill is installed unless runtime evidence shows it is installed.
• Do not require private memory, user profiles, Supabase, ContextHub, or any non-public context.
• Do not save tasks, write memory, install dependencies, update skills, or modify files unless the user explicitly asks for that action.

Source & license

This open-source skill is cataloged on AgentStack and links to its original source — we do not rehost the code.

• Author: fabioespindula
• Source: fabioespindula/awesome-nanoclaw-skills
• License: MIT
• Homepage: https://awesome-nanoclaw-skills.vercel.app

Install and usage instructions live in the source repository linked above.