stream-builder

The framework-agnostic Stream builder (Steps 0-7 scaffold + enhance + Video audit), being extended to support app kinds beyond web. Use ONLY when the user explicitly names 'stream-builder' or '/stream-builder'. Web React / Next.js work - 'build me a chat app', 'scaffold', 'create a new ...', 'add Chat/Video/Feeds to this app', 'integrate Video', livestreaming, video conferencing, team messaging, direct messaging, social feed - defaults to the stream-react skill, not this one. Not auto-selected for generic web build requests.

1.43x

Quality

81%

Does it follow best practices?

Impact

92%

1.43x

Average score across 3 eval scenarios

Securityby

Passed

No findings from the security scan

Quality

Content

85%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.

An exceptionally actionable, well-sequenced builder skill with verbatim commands, consent gates, and validation feedback loops throughout. Its one real defect is that two of its load-bearing references (builder-ui.md and enhance.md) do not exist in the bundle, which would stall the agent at Step 4 and Track E.

Suggestions

Add the missing `builder-ui.md` and `enhance.md` files to the bundle (or remove/redirect those links) — Step 4 makes `builder-ui.md` a mandatory load and Track E depends entirely on `enhance.md`, so both flows dead-end today.

Consolidate the Track F audit disclaimer into one place (e.g., the flows list) instead of repeating it in both the 'Start' section header and again in the flows; the repeated 3-sentence block costs tokens on every read.

Move the per-use-case Page Flow detail (livestream/conferencing/messaging/social-feed navigation trees) into a reference file like `references/PAGE-FLOW.md`, keeping only the hub-first principle and the key rules inline.

Dimension	Reasoning	Score
Conciseness	The body is dense with operational directives and spends almost no tokens explaining concepts Claude already knows — matching the 'efficient; minor instances of over-explanation' anchor. Not 5 because the Track F 'skips this' disclaimer is repeated three times and the Page Flow section carries reference-grade detail inline that could be trimmed or split out.	4 / 5
Actionability	Copy-paste-ready commands throughout: the scaffold one-liner with package-name fixup, `getstream env`, the SDK install matrix, the icon-presence check via `node -e`, `npx tsc --noEmit`, and the dev-server startup with a random port. Disclosure and theme-pick scripts are given verbatim, matching the 'fully executable; copy-paste ready' anchor.	5 / 5
Workflow Clarity	Steps 0-7 are explicitly sequenced with 'Execute phases in order' and 'Do not run independent phases in parallel', plus a numbered scaffold-order recap. Validation checkpoints are explicit: gitignore verification before writing secrets, tsc-first-then-build feedback loops with fix instructions, and stop-and-wait consent gates for the theme pick and third-party skill installs.	5 / 5
Progressive Disclosure	Product references and the use-case recipe manifest are one level deep, well signaled, and verified to exist, with load-only-relevant instructions. However, two files the body directs the agent to load — `builder-ui.md` (mandatory at Step 4) and `enhance.md` (the entire Track E) — are absent from the bundle, breaking navigation at core steps. This places it between 'some structure' and 'good structure', closer to the former because the missing files are load-bearing.	3 / 5
	Total	17 / 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 strong, conflict-aware description that explicitly separates this skill from its sibling stream-react skill and states precise invocation conditions. Its main weakness is that the capability statement is terse and jargon-heavy, requiring the body to understand what 'Steps 0-7 scaffold' actually does.

Dimension	Reasoning	Score
Specificity	Names the domain ('framework-agnostic Stream builder') and several concrete actions ('Steps 0-7 scaffold', 'enhance', 'Video audit'), matching the 'several specific actions; minor gaps' anchor. Not 5 because the actions are terse and jargon-bound ('Steps 0-7', 'Video audit' require the body to decode), and not 3 because multiple distinct capabilities are explicitly named.	4 / 5
Completeness	Both parts are present: the 'what' ('Stream builder (Steps 0-7 scaffold + enhance + Video audit)') and an explicit 'when' ('Use ONLY when the user explicitly names stream-builder or /stream-builder'). Not 5 because the 'what' leans on internal jargon and meta-commentary ('being extended to support app kinds beyond web') rather than plainly stating what it builds.	4 / 5
Trigger Term Quality	Includes many natural phrases users actually say ('build me a chat app', 'scaffold', 'livestreaming', 'video conferencing', 'team messaging', 'direct messaging', 'social feed') plus the explicit '/stream-builder' trigger — good coverage. Not 5 because most of those natural phrases are framed as anti-triggers that route to a different skill (stream-react), so the coverage is for disambiguation rather than invocation, and a few common synonyms are absent.	4 / 5
Distinctiveness Conflict Risk	Clear niche with explicit conflict avoidance: it names the competing skill ('defaults to the stream-react skill, not this one') and states 'Not auto-selected for generic web build requests.' Minimal risk of triggering for the wrong skill.	5 / 5
	Total	17 / 20 Passed

Validation

75%

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 — 12 / 16 Passed

Validation for skill structure

Criteria	Description	Result
allowed_tools_field	'allowed-tools' contains unusual tool name(s)	Warning
metadata_version	'metadata.version' is missing	Warning
relative_links	Relative link issues: 2 missing, 1 deeper-than-1-level, 3 suspicious	Warning
referenced_paths_exist	Referenced path issues: 2 deeper-than-1-level	Warning

	Total	12 / 16 Passed

Repository: GetStream/agent-skills
Path: skills/stream-builder/SKILL.md
Commit: 233f9cb

Reviewed: 17 hours 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.