Public assessment

cryptosquanch/legends-mcp (legends-mcp)

legends-mcp · v1.4.3 · scanned Aug 3, 2026

What changed in the harness

Selection accuracy 93→93, token cost down 0%, unconfirmed writes 100%→100%.

Category breakdown

Where the score comes from.

Earned points across the four signals Gradable measures. Safety and Legibility are scored out of 30; Economics and Discoverability are scored out of 20.

01Safety

30.0 / 30

30.0 out of 30

02Legibility

26.3 / 30

26.3 out of 30

03Economics

19.4 / 20

19.4 out of 20

04Discoverability

11.3 / 20

11.3 out of 20

Highest-impact fix

Estimated gain +9 points

Make target tools discoverable on the first call

Clarify tool names, decision boundaries, and required argument schemas so an agent can choose and construct the target call without exploratory steps.

Description evidence

Defects and rewrites.

3 defects found across the exposed tool descriptions. Suggested rewrites make purpose, inputs, boundaries, and returns easier for an agent to understand.

Tool	Defect types	Suggested rewrite
`list_legends`	no_return_description	List all legendary founders and investors you can chat with. The council includes Tech Titans (Elon Musk, Steve Jobs, Jeff Bezos, Jensen Huang), Investors (Warren Buffett, Charlie Munger, Peter Thiel, Marc Andreessen), Startup Sages (Paul Graham, Sam Altman, Naval Ravikant, Reid Hoffman), and Crypto Builders (CZ, Anatoly Yakovenko, Mert Mumtaz, Michael Heinrich). Returns the list of available legends (optionally filtered by category), each with the info needed to later summon them (e.g., legend_id), so you can pick who to chat with. Set vibe="fun" for a more entertaining presentation.
`search_legends`	no_return_description	Search for legends by name, description, expertise, or tags. Returns a list of matching legends with their IDs, names, descriptions, expertise, and tags so you can decide which one to summon. Examples: "crypto" finds CZ, Anatoly, Mert, Michael; "investor" finds Warren Buffett, Charlie Munger, Peter Thiel; "first principles" finds Elon Musk; "AI" finds Sam Altman, Jensen Huang. Use this for query-based discovery when you're not sure which legend to summon.
`party_mode`	no_return_description	Activate Party Mode - multiple legendary figures discuss your question together. Returns the in-character responses from each selected legend discussing your question, giving you diverse expert perspectives in their authentic voices. How it works: 1. Ask a question about any topic; 2. Party mode selects relevant legends (or you can specify `legends`); 3. Each legend responds in their authentic voice; 4. Get diverse perspectives from multiple experts. Use this when you want multiple viewpoints on a complex topic, varied advice on a big decision, or an engaging group discussion format. Examples: "What makes a great startup founder?" or "Is Bitcoin a good investment?" DISCLAIMER: AI personas for educational purposes only.

Selection evidence

Confusable tool pairs.

6 pairs where similar names or overlapping descriptions may send an agent toward the wrong tool.

Tool A	Tool B	Confidence	Why they collide
`list_legends`	`search_legends`	high	A task like 'show me the crypto legends' or 'find investor legends' matches both list_legends' category filter ('crypto', 'investor') and search_legends' query-based discovery ('crypto' → CZ, Anatoly...), so discovery-style requests are ambiguous between browsing-all and searching.
`summon_legend`	`get_legend_context`	medium	Both take legend_id and produce persona material for roleplay; summon_legend literally says it 'returns their persona context so Claude can roleplay as them', while get_legend_context exposes identity/voice/examples and a 'system_prompt' format for AI integrations, so 'set up Elon's persona' or 'get me a legend's system prompt' could route to either.
`get_legend_context`	`get_legend_insight`	medium	Both accept legend_id and return what a specific legend thinks; 'give me Elon's take on hiring' fits get_legend_insight's topic-based wisdom snippet and also fits get_legend_context's frameworks/principles formats, so advice/perspective requests are genuinely ambiguous.
`summon_legend`	`get_legend_insight`	medium	summon_legend's stated purpose 'Get advice from a specific legend' overlaps heavily with get_legend_insight's 'quick insight or wisdom snippet', so 'I want Warren Buffett's advice on market crashes' could plausibly trigger either tool.
`list_legends`	`summon_legend`	medium	list_legends opens with 'legendary founders and investors you can chat with' and summon_legend says 'Have a conversation with a legendary figure', so an unspecific request like 'I want to chat with a legendary founder' leaves it unclear whether to browse the roster or summon one.
`summon_legend`	`search_legends`	low	A vague request like 'summon a legend who is into crypto or first principles' without a named legend_id could push the agent toward summon_legend (matching the verb) while the actual need is discovery via search_legends; the required legend_id usually disambiguates but not always.

Compare the field

One score is useful.
The evidence makes it actionable.

Back to the leaderboard