Content
68%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.
A well-structured, executable skill body: concrete CLI commands, a useful action table, and a clear connection lifecycle with state polling. The main weaknesses are missing verification steps for destructive operations, a broken "Step 2" cross-reference, and interleaved generic Membrane boilerplate that dilutes Chatra-specific guidance.
Suggestions
Add a validation step for destructive and batch operations, e.g. "Before running delete-* actions, run the corresponding get-* action to confirm the target ID, and verify the result in the output field afterward".
Fix the cross-references: define or remove the "skip to **Step 2**" pointer and give the connection section consistent step numbering (1a/1b or Step 1/Step 2).
Standardize on the "membrane" invocation (explaining the one "npx @membranehq/cli" usage) and trim the generic Membrane setup boilerplate to tighten conciseness.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by concrete commands and tables with only minor over-explanation ("This is the fastest way to get a connection. The URL is normalized to a domain and matched against known apps...") and a grammar slip ("This will burn less tokens"). Every token mostly earns its place, but several boilerplate sentences could be trimmed to reach anchor 5. | 4 / 5 |
Actionability | Commands are copy-paste ready throughout ("membrane connection ensure \"https://chatra.io\" --json", "membrane action run <actionId> --connectionId=CONNECTION_ID --input '{\"key\": \"value\"}'") with a 10-action table and a proxy flag table. Minor gaps keep it below 5: it says "skip to **Step 2**" but no Step 2 is defined, and it alternates between "membrane" and "npx @membranehq/cli" without explanation. | 4 / 5 |
Workflow Clarity | The install → login → connect → poll → search → run sequence is present with state-polling checkpoints, but destructive actions (delete-pushed-message, delete-agent-message) ship with no validation or verification guidance, capping this dimension per the rubric. The dangling "Step 2" reference and an unnumbered "1b" section further blur the sequence. | 3 / 5 |
Progressive Disclosure | As a single-file skill with no bundle files, content is organized into clear one-level sections (install, auth, connect, search, actions, proxy, best practices) with no nested references. Minor gaps: the generic Membrane CLI setup could live in a shared reference, and the inconsistent step numbering ("1b" with no "1a", undefined "Step 2") slightly hinders navigation. | 4 / 5 |
Total | 15 / 20 Passed |