Advise Skills

When to use this skill

Use when the user wants to find skills relevant to a task or architectural decision. Invoked as /advise-skills (Claude Code) or @advise-skills (Cursor).

Configuration

• Primary source: All configured registries in ~/.prism/registries.json (with per-registry caching and source tagging)
• Fallback 1: Local skill-registry.json in the current project directory
• Fallback 2: Local _analysis/extracted_skills_codebase/ skill directories
• Authentication: REGISTRY_TOKEN environment variable (global override) or per-registry token from registries.json

> Skills from multiple registries are merged and each result is tagged with [registry-name] (e.g., [team], [community]) to identify the source.

Instructions

Step 1 -- Resolve skill registry source

Read ~/.prism/skills/_shared/registry-fetch.md and follow its instructions. It resolves all configured registries with ETag caching and local fallbacks, returning a skills array where each entry is tagged with _registry.

Outcomes:

• Valid JSON array → use as the skills list, proceed to Step 2
• REGISTRY_EMPTY → registry reached but empty, proceed to Step 2
• NO_REGISTRIES → no registry reachable, stop (the shared file handles the user message)

Step 2 -- Check for empty registry

If the skills array from the resolved source is empty, respond with: "No skills have been published to the registry yet." and stop.

Step 3 -- Extract trigger clauses

For each entry in the skills array, extract the TRIGGER when clause from its description field. This clause defines the scenarios where the skill is relevant.

Step 4 -- Match and rank

Compare the user's query against each trigger clause semantically. A skill matches if the user's intent aligns with at least one trigger scenario. Require genuine semantic alignment -- do not match on superficial keyword overlap.

Rank matches by relevance to the user's query.

Step 5 -- Respond with match table

Respond with a table using only registry metadata (no fetching yet). If the skill has a _registry field, prefix the skill name with [registry-name] to show the source:

| Skill Name | Repository | Why it matches | |---|---|---| | [registry-name] {name} | {repository} | {why it matches this query} |

If no skills match the query, state that clearly and list all available skill names with their one-line descriptions so the user can see what is available and refine their query. Stop here.

Then ask:

> "Fetch full skill details for all N match(es), or would you like to adjust the list first?"

Wait for the user's response. If they want to adjust, accept a revised list of skill names and proceed with those. If they confirm all, proceed with the full matched set.

Step 6 -- Load confirmed skills' SKILL.md

For each confirmed skill, try to load its full SKILL.md content.

If the skill was loaded from a remote registry: Fetch the individual skill's SKILL.md using the path field from the registry entry:

python3 /dev/null | sed 's/.*\///;s/\.git$//'
date +%d-%m-%Y

# /advise-skills (Claude Code) or @advise-skills (Cursor) Results
# {repository} -- {DD-MM-YYYY}

> Query: {the user's original query}

| Skill Name | Repository | Why it matches |
|---|---|---|
| {name} | {repository} | {why it matches this query} |

If no skills were successfully loaded, do not write the file.

Output rules

• Only respond with a table. Do not render the full SKILL.md files below the table.
• Do not add commentary or analysis beyond what is in the SKILL.md.
• Do not suggest skills that do not have a clear trigger match.
• Order results by relevance, strongest match first.

Source & license

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

• Author: ProsusAI
• Source: ProsusAI/prism
• License: Apache-2.0

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