delegate-to-agent

How to delegate all AI work to the agent chat. Use when delegating AI work from UI or scripts to the agent, when a user asks for agent behavior or LLM-powered features, when tempted to add inline LLM calls, or when sending messages to the agent from application code.

Quality

79%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Low

Low-risk findings worth noting

Fix and improve this skill with Tessl

tessl review fix ./.agents/skills/delegate-to-agent/SKILL.md

Quality

Content

75%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.

A highly actionable, well-structured policy skill: every delegation surface has executable code, decision tables resolve the main ambiguities, and risky auto-submit operations get explicit guardrails. Its weakness is redundancy — the completeText and auto-submit rules each appear in two or three sections — and the absence of any failure-path guidance keeps workflow clarity at 4.

Suggestions

State the completeText() escape hatch once (in 'How' or 'Exception') and have the 'Don't' entry reference it in one line, cutting roughly two repeated sections of explanation.

Merge the auto-submit guidance ('Capture user input in the sidebar' and the 'Don't auto-submit a hardcoded prompt' bullet) into a single rule with its examples to remove the duplicated rationale.

Add a short failure-path note for background runs and sub-agent delegation (e.g. how to detect and retry a failed run alongside useAgentChatGenerating) to close the workflow-clarity gap.

Dimension	Reasoning	Score
Conciseness	Mostly efficient app-specific guidance Claude could not infer (sendToAgentChat semantics, the message/context boundary), but the completeText escape-hatch rule is stated three times ('Rule', 'Don't', 'Exception') and the auto-submit rule twice ('Capture user input', 'Don't'), so the document could be meaningfully tightened. It is not anchor 4's 'minor instances' — the redundancy spans whole sections.	3 / 5
Actionability	Copy-paste-ready TypeScript for every surface (client, background, Node scripts, server transform, generating-state hook) plus two decision tables (message/context/images surfaces, submit true/false/omitted) and a concrete Button example. The common delegation cases are covered with executable code and specific option values.	5 / 5
Workflow Clarity	Decision flows are explicit and sequenced: surface selection by caller type, a submit-vs-prefill decision table, a sub-agent briefing contract (Objective/Context/Output/Boundaries) with fan-out and synthesis discipline, and checkpoints on risky operations ('Never auto-submit a generic creative prompt', prefilling 'Delete all projects older than 30 days' for user review). Falls short of anchor 5 because there are no error-recovery loops (e.g. what to do when callAgent or a sub-agent run fails).	4 / 5
Progressive Disclosure	No bundle files exist, and the body is well-sectioned with clear headers and one clearly signaled cross-skill reference ('See the a2a-protocol skill for the full pattern'). Minor gap: the ~50-line sub-agent teams section is a separable concern that could live in its own reference file. Not anchor 3 — structure and signaling are good, and inline placement of a 285-line policy document is defensible without a bundle.	4 / 5
	Total	16 / 20 Passed

Description

78%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 well-formed description with an explicit what and four concrete, naturally phrased trigger conditions. Its main weakness is capability specificity: it states one action (delegate to the agent chat) rather than enumerating the several distinct behaviors the body actually governs (sidebar capture, background runs, sub-agent delegation, A2A).

Suggestions

Enumerate 2-3 concrete capabilities after the 'what' clause (e.g. 'send prompts from UI or scripts to the agent chat, manage visible vs hidden context, delegate sub-agent work') to lift specificity from anchor 3 to 4.

Add a distinguishing trigger that separates it from sibling skills, e.g. 'for the local agent only — use a2a-protocol for remote agents', to reduce overlap risk on the broad 'agent behavior' phrase.

Include one or two synonym trigger terms users would naturally say, such as 'AI feature' or 'prompt the model', to round out trigger coverage.

Dimension	Reasoning	Score
Specificity	The description names its domain ('delegate all AI work to the agent chat') and one concrete capability (delegating from UI or scripts), but unlike anchor 4 it does not list several specific actions — the rest of the text is trigger conditions rather than capabilities.	3 / 5
Completeness	'How to delegate all AI work to the agent chat' answers the what, and four explicit trigger scenarios ('when delegating AI work from UI or scripts... when a user asks for agent behavior or LLM-powered features, when tempted to add inline LLM calls, or when sending messages to the agent') answer the when with concrete phrases, matching the anchor 5 example structure.	5 / 5
Trigger Term Quality	Natural developer phrases like 'delegating AI work from UI or scripts', 'LLM-powered features', 'inline LLM calls', and 'sending messages to the agent from application code' cover the main entry points well. It falls short of anchor 5's comprehensive synonym coverage (e.g. 'AI feature', 'call the model directly', specific API names).	4 / 5
Distinctiveness Conflict Risk	The inline-LLM-call and agent-chat framing carves out a distinct niche, but 'when a user asks for agent behavior or LLM-powered features' is broad enough to overlap with closely related skills (a2a-protocol, actions) mentioned in the body. It does not reach anchor 5's minimal-conflict distinct triggers.	4 / 5
	Total	16 / 20 Passed

Validation

81%

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 — 13 / 16 Passed

Validation for skill structure

Criteria	Description	Result
metadata_version	'metadata.version' is missing	Warning
metadata_field	'metadata' should map string keys to string values	Warning
frontmatter_unknown_keys	Unknown frontmatter key(s) found; consider removing or moving to metadata	Warning

	Total	13 / 16 Passed

Repository: BuilderIO/agent-native
Path: .agents/skills/delegate-to-agent/SKILL.md
Commit: a941a2e

Reviewed: 2 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.