Content
82%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 tight, example-driven reference for the batch command: fully executable guidance across CLI, config-file, and programmatic access, with genuinely useful pitfalls and error-recovery notes. The only gaps are mild redundancy, a missing retry loop for failed files, and reference paths that cannot be verified against the bundle.
Suggestions
Add an explicit retry/re-validation loop in Error recovery (e.g., re-run `xberg batch` on just the paths whose entries carry errors) to complete the validate -> fix -> retry pattern the batch context calls for.
Deduplicate the memory-cap advice (Parallelism vs Error recovery vs Pitfalls) into one place and cross-reference it, trimming the repeated `--format json` default note similarly.
Make the `references/cli-reference.md` and sibling-skill API references resolvable — either ship the file in a references/ directory or state its exact relative path so the pointers aren't dangling.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | Every section carries tool-specific knowledge Claude cannot infer (CLI defaults, flag semantics, JSON/TOML/Python examples) with no generic PDF or library explanation. Minor repetition keeps it from a 5 — the Error recovery section restates the Parallelism memory advice, and the format-default pitfall repeats the Basic usage note. | 4 / 5 |
Actionability | All examples are copy-paste executable and cover the common cases: bash glob batches, `--max-concurrent`/`--max-threads` caps, a concrete `--file-configs` JSON file, image-output setup with `mkdir -p`, a TOML config file, and a runnable Python `extract_batch` snippet. Not below 5 — no pseudocode or missing key details anywhere. | 5 / 5 |
Workflow Clarity | The batch-operation validation cap is satisfied: 'Inspect results for partial content and surfaced errors rather than relying on the process exit code alone' is an explicit checkpoint, and the pitfalls section flags concrete failure modes (--output-dir must exist, path-key matching). Held at 4 rather than 5 because no retry/re-validation loop is described for failed files within a batch. | 4 / 5 |
Progressive Disclosure | Sections are well organized and detail is pushed out ('See `references/cli-reference.md` for the full `batch` flag set'), matching the anchor-4 example. It is not a 5 because the referenced `references/cli-reference.md` does not exist in this skill's bundle (no references/ directory at all), and the Python/Node/Rust API references point into a 'sibling xberg skill' without a resolvable path. | 4 / 5 |
Total | 17 / 20 Passed |