Content
71%Weight 40%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
The body is a strong operational guide: every workflow step comes with an executable command, and connection-state error handling forms a genuine feedback loop. Its weaknesses are padding (concept-explaining intro, a description-free action table), a dangling "Step 2" reference, and no separation of catalog material into reference files.
Suggestions
Delete or trim the introductory paragraph explaining what Appwrite is, and drop the "No description" column or move the whole Popular actions table to a references/actions.md file with real descriptions.
Fix the broken navigation: "skip to **Step 2**" points to a step that is never defined — either label the sections as steps or rephrase to "skip to Searching for actions".
Add a brief post-run validation note (e.g. check the output field and HTTP/error status before reporting success), since actions include batch-style create/list operations on live project data.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The opening paragraph ("Appwrite is an open-source, self-hosted platform... abstracts away the complexities of backend development") explains a concept Claude already knows, the 19-row Popular actions table contributes nothing since every Description cell reads "No description", and filler like "Use action names and parameters as needed." adds tokens without guidance. The rest is mostly efficient command documentation, so this sits at anchor 3 ('mostly efficient but includes some unnecessary explanation') rather than 2's 'several padded sections' dominating the doc. | 3 / 5 |
Actionability | Guidance is fully executable and copy-paste ready: "npm install -g @membranehq/cli@latest", "membrane login --tenant --clientName=<agentType>", "membrane connection ensure \"https://appwrite.io/\" --json", "membrane action run <actionId> --connectionId=CONNECTION_ID --input '{\"key\": \"value\"}' --json", plus a concrete proxy flags table. Specific examples cover the common cases, matching the top anchor. | 5 / 5 |
Workflow Clarity | There is a clear install → login → connect → poll-state → search → run → proxy-fallback sequence, and the connection-state section includes explicit error-recovery guidance (CLIENT_ACTION_REQUIRED actions, polling again after user action, checking the error field on CONFIGURATION_ERROR). It falls short of 5 because "skip to **Step 2**" references a step label that is never defined anywhere in the document, and running actions has no validation guidance for verifying outputs. | 4 / 5 |
Progressive Disclosure | No bundle files exist, and catalog-style content is inlined directly in SKILL.md: the 19-row action catalog (all "No description") and the proxy flag reference table would fit better in a separate reference file. Section headers are well-organized, but 'content that should be separate is inline' places this at anchor 3 rather than 4's 'most content is appropriately placed'. | 3 / 5 |
Total | 15 / 20 Passed |