MCP servers / @perplexity-ai/mcp-server

MCP tool changes

@perplexity-ai/mcp-server

4 of its last 5 updates changed what the agent reads: 6 tool descriptions rewritten, up to 1.3.0.

npm package · npm page · recorded on 24 October 2026 on a clean machine with placeholder settings; only tools/list was sent. A changed description is not a vulnerability. Method and limits.

Release by release

1.2.1 → 1.3.0

minor release, 25 September 2026

perplexity_search: description and input schema changed; new arguments: search_type

Before (1.2.1)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. Supports recency filters and domain restrictions. For AI-generated answers with citations, use perplexity_ask instead.

After (1.3.0)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. Supports recency filters, domain restrictions, and a lower-latency fast search mode. For AI-generated answers with citations, use perplexity_ask instead.

1.2.0 → 1.2.1

patch release, 27 August 2026

perplexity_ask: input schema changed

perplexity_reason: input schema changed

perplexity_research: input schema changed

perplexity_search: input schema changed

1.1.0 → 1.2.0

minor release, 27 July 2026

No tool definition changed.

1.0.0 → 1.1.0

minor release, 27 July 2026

perplexity_search: description and input schema changed; new arguments: search_domain_filter, search_recency_filter

Before (1.0.0)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. For AI-generated answers with citations, use perplexity_ask instead.

After (1.1.0)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. Supports recency filters and domain restrictions. For AI-generated answers with citations, use perplexity_ask instead.

0.9.0 → 1.0.0

major release, 27 July 2026

perplexity_ask: description and input schema changed

Before (0.9.0)

Answer a question using web-grounded AI (Sonar Pro model). Best for: quick factual questions, summaries, explanations, and general Q&A. Returns a text response with numbered citations. Fastest and cheapest option. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For in-depth multi-source research, use perplexity_research instead. For step-by-step reasoning and analysis, use perplexity_reason instead.

After (1.0.0)

Answer a question using web-grounded AI (Perplexity Agent API, fast preset). Best for: quick factual questions, summaries, explanations, and general Q&A. Returns a text response with numbered citations. Fastest and cheapest option. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For in-depth multi-source research, use perplexity_research instead. For step-by-step reasoning and analysis, use perplexity_reason instead.

perplexity_reason: description and input schema changed; arguments removed: strip_thinking

Before (0.9.0)

Analyze a question using step-by-step reasoning with web grounding (Sonar Reasoning Pro model). Best for: math, logic, comparisons, complex arguments, and tasks requiring chain-of-thought. Returns a reasoned response with numbered citations. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For quick factual questions, use perplexity_ask instead. For comprehensive multi-source research, use perplexity_research instead.

After (1.0.0)

Analyze a question using step-by-step reasoning with web grounding (Perplexity Agent API, medium preset). Best for: math, logic, comparisons, complex arguments, and tasks requiring chain-of-thought. Returns a reasoned response with numbered citations. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For quick factual questions, use perplexity_ask instead. For comprehensive multi-source research, use perplexity_research instead.

perplexity_research: description and input schema changed; arguments removed: reasoning_effort, strip_thinking

Before (0.9.0)

Conduct deep, multi-source research on a topic (Sonar Deep Research model). Best for: literature reviews, comprehensive overviews, investigative queries needing many sources. Returns a detailed response with numbered citations. Significantly slower than other tools (30+ seconds). For quick factual questions, use perplexity_ask instead. For logical analysis and reasoning, use perplexity_reason instead.

After (1.0.0)

Conduct deep, multi-source research on a topic (Perplexity Agent API, high preset). Best for: literature reviews, comprehensive overviews, investigative queries needing many sources. Returns a detailed response with numbered citations. Significantly slower than other tools (can take minutes). For quick factual questions, use perplexity_ask instead. For logical analysis and reasoning, use perplexity_reason instead.

perplexity_search: description changed

Before (0.9.0)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) — no AI synthesis. For AI-generated answers with citations, use perplexity_ask instead.

After (1.0.0)

Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. For AI-generated answers with citations, use perplexity_ask instead.

Releases tried: 0.9.0 (2026-03-23); 1.0.0 (2026-07-27); 1.1.0 (2026-07-27); 1.2.0 (2026-07-27); 1.2.1 (2026-08-27); 1.3.0 (2026-09-25).

Tools in 1.3.0

The pin is the SHA-256 of the tool's name, description and input schema, as verfy.lock records it. The same pin in your lock means your agent reads this exact definition.

Tools in 1.3.0
Tool	Description	Pin
perplexity_ask	Answer a question using web-grounded AI (Perplexity Agent API, fast preset). Best for: quick factual questions, summaries, explanations, and general Q&A. Returns a text response with numbered citations. Fastest and cheapest option. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For in-depth multi-source research, use perplexity_research instead. For step-by-step reasoning and analysis, use perplexity_reason instead.	b6f208ac5008
perplexity_research	Conduct deep, multi-source research on a topic (Perplexity Agent API, high preset). Best for: literature reviews, comprehensive overviews, investigative queries needing many sources. Returns a detailed response with numbered citations. Significantly slower than other tools (can take minutes). For quick factual questions, use perplexity_ask instead. For logical analysis and reasoning, use perplexity_reason instead.	b01f29c9001e
perplexity_reason	Analyze a question using step-by-step reasoning with web grounding (Perplexity Agent API, medium preset). Best for: math, logic, comparisons, complex arguments, and tasks requiring chain-of-thought. Returns a reasoned response with numbered citations. Supports filtering by recency (hour/day/week/month/year), domain restrictions, and search context size. For quick factual questions, use perplexity_ask instead. For comprehensive multi-source research, use perplexity_research instead.	9d6aa0b7cac2
perplexity_search	Search the web and return a ranked list of results with titles, URLs, snippets, and dates. Best for: finding specific URLs, checking recent news, verifying facts, discovering sources. Returns formatted results (title, URL, snippet, date) with no AI synthesis. Supports recency filters, domain restrictions, and a lower-latency fast search mode. For AI-generated answers with citations, use perplexity_ask instead.	63c18fccad60

Lock these tools

Pin what @perplexity-ai/mcp-server serves you today, commit the lock, and a pull request fails when a tool changes, with the text before and after. Pin the package version in your MCP config as well.

$ pip install verfy
$ verfy lock check --update   # shows each tool, then writes verfy.lock
$ git add verfy.lock

# .github/workflows/verfy.yml, after actions/checkout
- uses: anvarbekovjavohir277-dot/verfy-action@v1

Maintain this server? Run the same check on your releases and add the Locked by Verfy badge to your README.