Tier-2 Review: render-deploy (crypto-web3 cluster)

What We Attempted

We evaluated the render-deploy skill, which is designed to guide an AI assistant in helping users deploy applications to Render's cloud platform. The skill provides two primary workflows: generating render.yaml Blueprint files for Infrastructure-as-Code deployments, and using MCP tools for direct service creation. It also includes decision heuristics for choosing between these methods based on the complexity of the user's application.

What Failed

In our test harness, the skill achieved 0 clean passes, 1 partial pass, and 1 failure:

1. Install (partial): The skill is a prompt/guide rather than an installable package, so there is no single "install" command for the skill itself. The documented prerequisite—installing the Render CLI via brew or curl—is a dependency that can fail if those tools are not available in the environment. This is acceptable for a guide-based skill, but the partial mark reflects the lack of a self-contained installation path.

2. Smoke-invocation (fail): The skill requires two prerequisites that are not satisfied in a clean container: a Git remote (GitHub, GitLab, or Bitbucket) and a Render API key (for MCP or CLI authentication). A minimal invocation like list_services() fails immediately because the Render CLI is either not installed or not authenticated, and MCP is not configured. The skill's own instructions state that if there is no Git remote, the assistant should stop and ask the user to create one—but this blocking condition is not handled gracefully in an automated test environment.

What We Observed

The core failure mode is unmet external prerequisites. The skill is well-written for human interaction but assumes a pre-configured environment that our clean container does not provide. Specifically:

• Render API key is required for any CLI or MCP operation. The skill documents that MCP setup is needed but does not include steps to obtain or configure the API key within the skill itself.
• Git remote is a hard requirement for both Blueprint and Direct Creation methods. The skill correctly identifies this as a blocking condition, but in an automated test, this results in a failure rather than a graceful fallback.
• Render CLI installation depends on brew or curl, which may not be present in minimal containers.

Rating Acknowledgment

The composite score of 4.3/5.0 is theoretical until a physical re-run can resolve the prerequisite failures. The dimensions that were scored (trigger clarity, output specificity, scope precision, self-containment) reflect the skill's design quality, not its operational reliability. The reusability score of 3.5 is arguably generous given the dependency on external services and credentials.

Conclusion

Despite the test failures, this skill remains valuable in principle. It provides a structured, well-documented workflow for deploying to Render, with clear decision rules for choosing between Blueprint and Direct Creation methods. The skill's logic—analyzing codebases, generating render.yaml files, and guiding users through deployment—is sound and would be useful in a real-world scenario where the prerequisites are met. The failures we observed are environmental, not logical.

However, the skill would benefit from explicit prerequisite handling: either a fallback path for environments without a Git remote or API key, or a clear "setup" section that can be executed before the main workflow. As it stands, the skill is a guide for interactive use, not a self-contained automation.

Tests simulated against README claims; pending physical re-run in Docker harness. Ran 2026-06-08.

Overall: broken. 0 tests passed, 1 partial, 1 failed; key blocker: missing Render API key and Git remote, which are prerequisites not satisfied in a clean container.

Inferred dependencies: git, render-cli (optional, via brew or curl), Render API key (for MCP), Git remote (GitHub/GitLab/Bitbucket).

Test	Status	Notes
install	partial	SKILL.md documents installing Render CLI via brew or curl, but does not specify a single 'install' command for the skill itself. The skill is a prompt/guide, not a package. Installation of Render CLI may fail if curl/brew not available.
smoke-invocation	fail	SKILL.md requires a Git remote and Render API key for MCP or CLI. Minimal invocation like 'list_services()' will fail without MCP setup or Render CLI installed and authenticated. No API key or Git repo is present in a clean container.