Content
75%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 strong, command-first CLI guide: nearly every instruction is an executable command, the connection lifecycle is handled with explicit state checkpoints and recovery paths, and there is almost no padding or explanation of concepts Claude already knows. The main defects are small but real — broken cross-references to a nonexistent 'Step 2' and unlabeled step '1b', a duplicated action-list command, and a login invocation with an unresolved `--tenant` placeholder.
Suggestions
Fix the internal navigation: either create a labeled "Step 2" section (e.g. Searching for actions) or remove the two "Skip to **Step 2**" references and rename "1b" to something coherent like "Wait for the connection to be ready".
Merge "Popular actions" into "Searching for actions" — it currently restates the same `action list --intent=QUERY` command with no added content.
Complete the login example: show what `--tenant` expects (e.g. `--tenant=<tenant-id>`) and link the `--clientName` value to the Agent Types list so the command is truly copy-paste ready.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is dominated by lean, executable commands with minimal concept explanation — it assumes competence rather than teaching CLI basics. Minor trimming is possible: the action-list command appears twice ("Searching for actions" and "Popular actions"), and asides like "This is the fastest way to get a connection" plus the vague "Agent Types" line add mild padding — so anchor 4 ('efficient; minor instances that could be trimmed'), not 5. | 4 / 5 |
Actionability | Nearly everything is copy-paste ready: `npm install -g @membranehq/cli@latest`, `membrane connection ensure "https://www.clarify.io/" --json`, `membrane action run <actionId> --input '{...}' --json`, plus a concrete proxy flags table. Not 5 because of small gaps: "Skip to **Step 2**" points to a section that doesn't exist, and `membrane login --tenant --clientName=<agentType>` leaves `--tenant` without a value or placeholder explanation. | 4 / 5 |
Workflow Clarity | The install → login → connect → poll → search → run flow is clearly sequenced with explicit state-based checkpoints (READY / BUILDING / CLIENT_ACTION_REQUIRED / CONFIGURATION_ERROR) and a re-poll loop after user action — genuine error-recovery feedback. Not 5 because the section numbering is incoherent ("1b" with no labeled Step 1, and two references to a nonexistent "Step 2"), which leaves a navigation gap in the sequence. | 4 / 5 |
Progressive Disclosure | The skill is a single self-contained file with no references/, scripts/, or assets/ bundle, and the body contains no dangling references to bundle files. Sections (Install, Authentication, Connecting, Searching, Running, Proxy, Best practices) are well organized and appropriately sized for one level. Not 5 because at ~145 lines the proxy flag table and connection-state reference sit inline where a split into a reference file would keep the overview tighter. | 4 / 5 |
Total | 16 / 20 Passed |