intent-recognition

Classifies automation requests using two decisions: anchor (which primitive owns the top-level control flow — workflow-anchored, agent-anchored, needs-clarification, or out-of-scope) and embeds_other (whether the other primitive appears embedded inside — an agent step inside a workflow, or a workflow invoked as an agent tool). Must be used before deciding the intent of any automation request, including compound requests with multiple independent automations, mid-build extensions to an existing workflow or agent, one-off questions or reports that need external systems you cannot query directly, and requests that need clarification before an anchor can be chosen, before choosing workflow-builder, planning, or an agent-oriented design.

Quality

81%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Medium

Suggest reviewing before use

Quality

Content

77%Weight 40%Scale 1-3

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

A highly actionable, well-sequenced routing procedure with strong decision criteria and feedback loops. Its weaknesses are cross-section redundancy that hurts conciseness and a single-file structure with no progressive disclosure for a substantial skill.

Suggestions

Deduplicate the chat-trigger, cron-not-a-signal, and degenerate-shell rules so each appears once (likely in Gotchas) and is merely referenced elsewhere, cutting roughly a third of the body.

Move the worked Examples (and optionally the full Signals catalog) into a reference file under references/ and keep a concise representative subset inline, improving progressive disclosure for this long skill.

Dimension	Reasoning	Score
Conciseness	The body is dense with n8n-specific routing knowledge Claude does not already know and avoids basic-concept filler, but the same rules (the Chat Trigger + AI Agent node gotcha, cron-is-not-a-signal, the degenerate-shell check) are restated across Decisions, Signals, Gotchas, and Examples, which could be tightened by deduplication.	2 / 3
Actionability	It gives a concrete numbered decision procedure (steps 0–9), explicit criteria ('any one is enough' for agent signals, 'all must hold' for workflow), routing targets, and a copy-paste output format — fully actionable guidance for an instruction-only skill.	3 / 3
Workflow Clarity	The decision steps are clearly sequenced with explicit checkpoints and feedback loops: step 0 context continuity, step 7 degenerate-shell check to 're-run while building', and 'create prerequisites and call build-agent again' retry on missing workflows or tables.	3 / 3
Progressive Disclosure	Sections are well-organized with clear headers (Purpose, Decisions, Signals, Examples, Gotchas, Output Format), but the ~370-line body is monolithic with no bundle references, and content such as the 17 worked examples and detailed signals is inline where it could be split into a reference file.	2 / 3
	Total	10 / 12 Passed

Description

85%Weight 40%Scale 1-3

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 specific, complete, and distinct description that clearly states both what it does and when to use it. Its main weakness is heavy reliance on internal taxonomy jargon over the natural language users would actually invoke it with.

Suggestions

Add natural trigger phrasing users would actually say (e.g., 'Use when the user asks to build or automate something and you must decide between a workflow and an agent') alongside the jargon.

Reduce internal vocabulary in the lead sentence — lead with the user-facing problem ('deciding whether an automation should be a workflow or an agent') before introducing 'anchor' and 'embeds_other'.

Dimension	Reasoning	Score
Specificity	Names the concrete decisions ('anchor' with four enumerated values — workflow-anchored, agent-anchored, needs-clarification, out-of-scope — and 'embeds_other' with embedded-agent and workflow-as-tool examples), going beyond naming a domain to listing specific outcomes.	3 / 3
Completeness	It states what the skill does ('Classifies automation requests using two decisions') and gives an explicit when clause ('Must be used before deciding the intent of any automation request, including compound requests..., mid-build extensions...'), satisfying both what and when.	3 / 3
Trigger Term Quality	Some natural terms appear ('automation requests', 'one-off questions or reports', 'compound requests'), but the description is dominated by internal jargon ('anchor', 'embeds_other', 'primitive', 'workflow-anchored') that users would not naturally say, and it misses common variations like 'build a workflow' or 'automate X'.	2 / 3
Distinctiveness Conflict Risk	The intent-classification niche is distinct, and the description explicitly positions it before ('before choosing workflow-builder, planning, or an agent-oriented design'), reducing conflict with those adjacent skills.	3 / 3
	Total	11 / 12 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: n8n-io/n8n
Path: packages/@n8n/instance-ai/skills/intent-recognition/SKILL.md
Commit: bcbea0a

Reviewed: 7 days 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.