exa-search

Neural search via Exa MCP for web, code, and company research. Use when the user needs web search, code examples, company intel, people lookup, or AI-powered deep research with Exa's neural search engine.

1.55x

Quality

80%

Does it follow best practices?

Impact

93%

1.55x

Average score across 3 eval scenarios

Securityby

Low

Low-risk findings worth noting

Quality

Content

81%Weight 40%Scale 1-5

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

Highly actionable content: complete MCP setup and copy-paste-ready signatures for all six tools, with clear single-purpose workflow guidance. The weakness is redundancy — the Usage Patterns and Tips sections largely duplicate Core Tools — and inline bulk API tables where a reference file would help.

Suggestions

Cut the 'Usage Patterns' section or reduce it to one composite example; 'Quick Lookup' and 'Code Research' repeat calls already shown verbatim in Core Tools.

Merge 'Tips' into the tool sections (e.g. the tokensNum range guidance belongs in the get_code_context_exa notes column) to remove duplicated parameter guidance.

Move the six parameter tables into a references/tools.md file and keep one-line tool summaries in SKILL.md to improve progressive disclosure.

Dimension	Reasoning	Score
Conciseness	Mostly efficient — no explaining of concepts Claude already knows — but the 'Usage Patterns' section re-demonstrates tool calls already shown verbatim in 'Core Tools', and 'Tips' restates parameter guidance already in the tables. These are whole duplicated sections rather than minor instances, so it sits below the 'efficient; minor trimmable instances' anchor.	3 / 5
Actionability	The MCP config JSON (npx exa-mcp-server with env key) is copy-paste ready, and every one of the six tools has a complete executable call signature with a full parameter table including types, defaults, and constraints (e.g. tokensNum 1000-50000). No gaps remain for the common cases.	5 / 5
Workflow Clarity	This is a single-purpose skill (invoke Exa search tools) where the simple-skill exception applies: the single action is unambiguous via the 'When to Activate' triggers and the 'Tips' tool-selection guidance. The one multi-step pattern (deep_researcher_start → do other work → deep_researcher_check) is correctly sequenced with the polling step explicit, and there are no destructive or batch operations requiring validation.	5 / 5
Progressive Disclosure	No bundle files exist and the ~165-line body is cleanly sectioned (When to Activate, MCP Requirement, Core Tools, Usage Patterns, Tips, Related Skills) with clear navigation. It falls short of a 5 because the six inline parameter tables are bulk API reference that could live in a separate reference file.	4 / 5
	Total	17 / 20 Passed

Description

80%Weight 40%Scale 1-5

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

A strong description with a clear what/when structure and explicit trigger phrases. Its main weakness is distinctiveness: the generic 'web search' and 'deep research' triggers invite conflicts with built-in search and the deep-research skill.

Suggestions

Qualify the generic triggers to reduce conflicts, e.g. 'Use when the user asks for web search via Exa/neural search specifically' rather than bare 'web search'.

Drop or de-emphasize 'AI-powered deep research' in the description since the body's Related Skills already routes deep research to a separate skill; keeping it here creates overlap.

Add common user synonyms such as 'find', 'look up', or 'research companies' to the 'Use when' clause to improve trigger-term coverage toward a 5.

Dimension	Reasoning	Score
Specificity	Names several concrete capability areas — "neural search via Exa MCP for web, code, and company research" plus "code examples, company intel, people lookup, or AI-powered deep research" — with only minor gaps. Not a 5 because every action is a variation of 'search' rather than distinct fine-grained operations.	4 / 5
Completeness	Explicitly answers both: the 'what' ("Neural search via Exa MCP for web, code, and company research") and an enumerated 'when' ("Use when the user needs web search, code examples, company intel, people lookup, or AI-powered deep research...") with concrete trigger phrases.	5 / 5
Trigger Term Quality	Natural phrases like "web search", "code examples", "people lookup", and "deep research" are present, but common synonyms users would actually say ("find", "look up", "research companies", "news") are missing, so it falls just short of comprehensive synonym coverage.	4 / 5
Distinctiveness Conflict Risk	Naming Exa gives it a niche, but the generic triggers "web search" and "AI-powered deep research" overlap the built-in WebSearch capability and the related deep-research skill (which the body itself lists under Related Skills). It is somewhat specific but could still trigger the wrong skill.	3 / 5
	Total	16 / 20 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Repository: affaan-m/ECC
Path: .agents/skills/exa-search/SKILL.md
Commit: c70874f

Reviewed: 1 day ago

Table of Contents

Discovery Implementation Validation

Is this your skill?

If you maintain this skill, you can claim it as your own. Once claimed, you can manage eval scenarios, bundle related skills, attach documentation or rules, and ensure cross-agent compatibility.