Content
71%Scale 1-5Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.
This is a highly actionable, domain-expert skill that captures critical acli quirks and workarounds that Claude could not know independently. Its main weakness is length — at 300+ lines with repeated patterns (site-guard blocks, temp file explanations) and some sections that could be extracted to reference files, it consumes more tokens than necessary. The workflow structure is solid but could benefit from tighter error recovery loops for batch operations.
Suggestions
Extract the repeated site-guard snippet into a short macro reference (e.g., 'Run the site guard before every write command — see Auth section') instead of inlining the full block 7+ times.
Move the 'Known limitations' and 'View a work item' full extraction script into a reference file (e.g., `acli-reference.md`) to reduce the main skill's token footprint.
Remove the 'Why acli over MCP' section — the opening paragraph already establishes acli as the only transport; token cost rationale is unnecessary context for Claude.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The skill contains substantial domain-specific knowledge Claude wouldn't know (acli quirks, flag gotchas, ADF requirements, link direction semantics), but it's quite long (~300+ lines) with some redundancy — the site-guard block is repeated in nearly every code example, the 'Known limitations' section is verbose, and the 'Why acli over MCP' section explains rationale Claude doesn't need. The temp file pattern explanation is also repeated across sections. | 3 / 5 |
Actionability | Excellent actionability — every operation has copy-paste-ready bash commands with concrete flags, JSON schemas for bulk creation, jq pipelines for parsing output, and specific error messages to surface. The flag reference table, known limitations with workarounds, and counter-intuitive behavior callouts (--in/--out direction, --parent takes KEY not ID) are highly actionable. | 5 / 5 |
Workflow Clarity | The 'Workflow pattern for story creation' section provides a clear 7-step sequence. Auth guard with DNS check → auth → site guard is a well-structured validation chain. The verification steps for parent/label writes via JQL are good checkpoints. However, the workflow pattern is presented as a comment-style list rather than with explicit validation gates between steps, and there's no explicit error recovery loop for bulk creation failures. | 4 / 5 |
Progressive Disclosure | The skill references external files (`.claude/refs/jira-adf.md`, `scripts/jira-site-guard.sh`, `scripts/tmp-dir.sh`, `scripts/jira-set-field.sh`) which is good progressive disclosure, but the SKILL.md itself is monolithic — the known limitations, link semantics deep-dive, and view command's full extraction script could be split into reference files. The ADF reference is well-delegated, but the main file carries too much inline detail for its length. | 3 / 5 |
Total | 15 / 20 Passed |