Skill & Prompt Migration (v1 → v2)

How to migrate skills, prompts, and agent instructions from individual tools to the 5 meta-tools architecture

Overview

MCP v2 replaces 70+ individual tools with 5 universal meta-tools. Any skill, prompt, or agent instruction that references old tool names must be updated.

What changed:

v1 (old)	v2 (new)
`search_linkedin_users(keywords, location)`	`execute("linkedin", "search", "search_users", {"keywords": ..., "location": ...})`
70+ tool definitions in context	5 compact meta-tools
No caching	Built-in cache with `query_cache` and `export_data`
No server-side filtering	Server-side filtering, aggregation, grouping

Tool Mapping

Old tool	New call
`search_linkedin_users(keywords, location, count)`	`execute("linkedin", "search", "search_users", {"keywords": ..., "location": ..., "count": ...})`
`get_linkedin_profile(user)`	`execute("linkedin", "user", "get", {"user": ...})`
`get_linkedin_company(company)`	`execute("linkedin", "company", "get", {"company": ...})`
`search_linkedin_companies(keywords, count)`	`execute("linkedin", "search", "search_companies", {"keywords": ..., "count": ...})`
`search_linkedin_jobs(keywords, location, count)`	`execute("linkedin", "job_search", "search_jobs", {"keywords": ..., "count": ...})`
`search_linkedin_posts(keywords, count)`	`execute("linkedin", "post", "search_posts", {"keywords": ..., "count": ...})`
`get_linkedin_user_posts(user)`	`execute("linkedin", "post", "get_user_posts", {"user": ...})`
`find_linkedin_email(user)`	`execute("linkedin", "email", "find", {"user": ...})`
`google_linkedin_search(query, count)`	`execute("linkedin", "google", "search", {"query": ..., "count": ...})`

Twitter/X

Old tool	New call
`search_twitter_users(query)`	`execute("twitter", "search", "search_users", {"query": ...})`
`get_twitter_user(username)`	`execute("twitter", "user", "get", {"username": ...})`
`get_twitter_user_tweets(username)`	`execute("twitter", "user_tweets", "get", {"username": ...})`

Instagram

Old tool	New call
`search_instagram_users(query)`	`execute("instagram", "search", "search_users", {"query": ...})`
`get_instagram_user(username)`	`execute("instagram", "user", "get", {"username": ...})`
`get_instagram_post(url)`	`execute("instagram", "post", "get", {"url": ...})`

YouTube

Old tool	New call
`search_youtube(query, count)`	`execute("youtube", "search", "search_videos", {"query": ..., "count": ...})`
`get_youtube_channel(channel_id)`	`execute("youtube", "channel", "get", {"channel_id": ...})`
`get_youtube_video(video_id)`	`execute("youtube", "video", "get", {"video_id": ...})`

Old tool	New call
`search_reddit(query)`	`execute("reddit", "search", "search", {"query": ...})`
`get_reddit_user(username)`	`execute("reddit", "user", "get", {"username": ...})`
`get_reddit_posts(subreddit)`	`execute("reddit", "posts", "get", {"subreddit": ...})`

YC / SEC / Web

Old tool	New call
`search_yc_companies(query)`	`execute("yc", "search", "search", {"query": ...})`
`get_yc_company(slug)`	`execute("yc", "company", "get", {"slug": ...})`
`search_sec_filings(query)`	`execute("sec", "search", "search", {"query": ...})`
`get_sec_document(url)`	`execute("sec", "document", "get", {"url": ...})`
`scrape_webpage(url)`	`execute("webparser", "parse", "parse", {"url": ...})`

If an old tool name is not in the mapping above, use discover(source, category) to find the correct endpoint name, then call execute().

Migration Rules

1. Direct tool references → execute()

Before:

Use search_linkedin_users to find people matching the criteria.

After:

Use execute("linkedin", "search", "search_users", {params}) to find people matching the criteria.

2. Unknown endpoint/params → add discover() first

Before:

Search for the company on LinkedIn and get their details.

After:

Use discover("linkedin", "company") to check available endpoints and params.
Then use execute("linkedin", "company", "get", {params}) to get company details.

3. Known endpoint — skip discover()

discover() is only needed when the skill doesn't know the exact endpoint name or parameter format. If the skill hardcodes specific execute() calls, discover is not required.

4. Multi-step workflows

Before:

1. Use search_linkedin_users to find the person
2. Use get_linkedin_profile to get their full profile
3. Use find_linkedin_email to get their email

After:

1. Use execute("linkedin", "search", "search_users", {"first_name": ..., "last_name": ..., "count": 5})
   to find the person
2. Use execute("linkedin", "user", "get", {"user": "{alias from step 1}"})
   to get their full profile
3. Use execute("linkedin", "email", "find", {"user": "{alias from step 1}"})
   to get their email

5. New capabilities — pagination, filtering, export

v2 adds cache-based tools that didn't exist in v1. Update skills to take advantage:

Pagination (new)

Results from `execute()` include `cache_key`. If more data exists:
```
get_page(cache_key="{cache_key}", offset=10, limit=10)
```

Server-side filtering (new)

Filter results without consuming context tokens:
```
query_cache(cache_key="{cache_key}", conditions=[
  {"field": "location", "op": "contains", "value": "San Francisco"}
])
```

Aggregation (new)

Calculate statistics server-side:
```
query_cache(cache_key="{cache_key}", aggregate={"field": "followers", "op": "avg"}, group_by="industry")
```

Export to file (new)

Download full datasets:
```
export_data(cache_key="{cache_key}", output_format="csv")
→ returns download URL
```

6. Error handling

Before:

If search_linkedin_users returns an error, try with different keywords.

After:

If execute() returns an error with "llm_hint", follow the hint.
If execute() returns {"error": "Source not found", "available_sources": [...]}, check source name.
If execute() returns {"error": "Endpoint not found", "available_endpoints": [...]},
  call discover() to get correct endpoint names.

Migration Checklist

Replace tool calls

- Replace all `search_linkedin_*`, `get_linkedin_*` → `execute("linkedin", ...)`
- Replace all `search_twitter_*`, `get_twitter_*` → `execute("twitter", ...)`
- Replace all `search_instagram_*`, `get_instagram_*` → `execute("instagram", ...)`
- Replace all `search_youtube_*`, `get_youtube_*` → `execute("youtube", ...)`
- Replace all `search_reddit_*`, `get_reddit_*` → `execute("reddit", ...)`
- Replace all `search_yc_*`, `get_yc_*` → `execute("yc", ...)`
- Replace all `search_sec_*`, `get_sec_*` → `execute("sec", ...)`
- Replace all `scrape_webpage` → `execute("webparser", "parse", ...)`

Add new capabilities

- Add `get_page` for large result sets
- Add `query_cache` for filtering/aggregation
- Add `export_data` for file downloads

Add discover() where needed

Only where endpoint names or params are not known in advance

Clean up

- Remove references to disabled sources (e.g., Crunchbase)
- Update error handling to v2 format

Test

Test each migrated skill end-to-end

Automated Migration

With Claude Code Skill

Instead of migrating manually, use the dedicated anysite-mcp-migration skill for Claude Code:

Install via Claude Code:

# Add the skills marketplace
/plugin marketplace add https://github.com/anysiteio/agent-skills

# Install the migration skill
/plugin install anysite-mcp-migration@anysite-skills

Quick preview of all available skills:

npx @anysiteio/agent-skills

Usage — ask Claude Code in natural language:

"Migrate the skill at /path/to/SKILL.md to v2"
"Migrate this prompt to the new anysite MCP" (paste old prompt)
"What v1 tools are still in this skill?" (paste text)

The skill automatically:

Scans for all v1 tool references
Replaces with correct execute() calls
Adds discover() only where needed
Adds get_page, query_cache, export_data where beneficial
Removes disabled source references
Updates error handling
Outputs a migration summary

With Any LLM (Auto-Migration Prompt)

Copy the prompt below and paste it along with your old skill text into any LLM:

Auto-Migration Prompt (click to expand)

You are migrating an Anysite MCP skill from v1 (individual tools) to v2 (meta-tools).

RULES:
1. Replace every old tool call with execute(source, category, endpoint, params).
2. Only add discover() if the skill text says "check what's available" or doesn't specify
   exact endpoint/params. If the skill already knows exactly what to call — use execute()
   directly, no discover needed.
3. Add get_page/query_cache/export_data where the skill would benefit from pagination,
   filtering, or file export.
4. Keep the same logical flow — don't change what the skill does, only how it calls tools.
5. Remove references to Crunchbase (disabled source).

TOOL MAPPING:
- search_linkedin_users(...) → execute("linkedin", "search", "search_users", {...})
- get_linkedin_profile(user=X) → execute("linkedin", "user", "get", {"user": X})
- get_linkedin_company(company=X) → execute("linkedin", "company", "get", {"company": X})
- search_linkedin_companies(...) → execute("linkedin", "search", "search_companies", {...})
- search_linkedin_jobs(...) → execute("linkedin", "job_search", "search_jobs", {...})
- search_linkedin_posts(...) → execute("linkedin", "post", "search_posts", {...})
- get_linkedin_user_posts(user=X) → execute("linkedin", "post", "get_user_posts", {"user": X})
- find_linkedin_email(user=X) → execute("linkedin", "email", "find", {"user": X})
- google_linkedin_search(query=X) → execute("linkedin", "google", "search", {"query": X})
- search_twitter_users(query=X) → execute("twitter", "search", "search_users", {"query": X})
- get_twitter_user(username=X) → execute("twitter", "user", "get", {"username": X})
- get_twitter_user_tweets(username=X) → execute("twitter", "user_tweets", "get", {"username": X})
- search_instagram_users(query=X) → execute("instagram", "search", "search_users", {"query": X})
- get_instagram_user(username=X) → execute("instagram", "user", "get", {"username": X})
- search_youtube(query=X) → execute("youtube", "search", "search_videos", {"query": X})
- get_youtube_channel(channel_id=X) → execute("youtube", "channel", "get", {"channel_id": X})
- search_reddit(query=X) → execute("reddit", "search", "search", {"query": X})
- get_reddit_user(username=X) → execute("reddit", "user", "get", {"username": X})
- search_yc_companies(query=X) → execute("yc", "search", "search", {"query": X})
- get_yc_company(slug=X) → execute("yc", "company", "get", {"slug": X})
- search_sec_filings(query=X) → execute("sec", "search", "search", {"query": X})
- scrape_webpage(url=X) → execute("webparser", "parse", "parse", {"url": X})

NEW CAPABILITIES (add where useful):
- get_page(cache_key, offset, limit) — paginate large results
- query_cache(cache_key, conditions, sort_by, aggregate, group_by) — filter/sort/aggregate server-side
- export_data(cache_key, output_format) — download as json/csv/jsonl

If an old tool name is not in the mapping above, use discover(source, category) to find
the correct endpoint, then execute().

Now migrate the following skill:

<paste old skill text here>

Skill & Prompt Migration (v1 → v2)

Overview

Tool Mapping

LinkedIn

Twitter/X

Instagram

YouTube

Reddit

YC / SEC / Web

Migration Rules

1. Direct tool references → execute()

2. Unknown endpoint/params → add discover() first

3. Known endpoint — skip discover()

4. Multi-step workflows

5. New capabilities — pagination, filtering, export

Pagination (new)

Server-side filtering (new)

Aggregation (new)

Export to file (new)

6. Error handling

Migration Checklist

Replace tool calls

Add new capabilities

Add discover() where needed

Clean up

Test

Automated Migration

With Claude Code Skill

With Any LLM (Auto-Migration Prompt)

Auto-Migration Prompt (click to expand)