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 highly actionable, with executable gdb-cli commands throughout, a clearly sequenced five-step workflow, and useful indicator-to-investigation patterns. Its weaknesses are token efficiency — a duplicative Examples section, a When-to-Use list that repeats the description, and inline version pins — and the absence of any progressive disclosure, with patterns and examples that belong in reference files inlined in a 200+ line monolith.
Suggestions
Delete or merge the 'Examples' section — its commands duplicate 'How It Works' and 'Common Debugging Patterns' almost verbatim.
Move 'Common Debugging Patterns' and the worked correlation example into a references/ file (e.g. PATTERNS.md) linked one level deep from SKILL.md to enable progressive disclosure.
Trim 'When to Use This Skill' to trigger phrases that differ from the frontmatter description, and note version requirements (GDB 9.0+, Python 3.6.8+) in a dedicated prerequisites line rather than repeating them in prose.
| Dimension | Reasoning | Score |
|---|---|---|
Conciseness | The body is mostly efficient — commands are shown once with flags and brief comments — but the 'Examples' section largely re-runs commands already demonstrated in 'How It Works' (e.g. 'gdb-cli load --core', 'gdb-cli thread-apply -s $SESSION bt --all'), 'When to Use This Skill' duplicates the description, and pinned version numbers ('GDB 9.0+', 'Python 3.6.8+') sit outside any old-patterns section. This matches 'mostly efficient but includes some unnecessary explanation or could be tightened' rather than the minor-trim level of 4. | 3 / 5 |
Actionability | Every step is a copy-paste-ready gdb-cli invocation with flags and a concrete session_id convention ('"session_id": "a1b2c3"'), patterns pair indicators with exact investigation commands, and worked examples cover both common cases (core dump load and live attach). This matches 'fully executable; copy-paste ready code or commands; specific examples cover the common cases'. | 5 / 5 |
Workflow Clarity | The five numbered steps (initialize → gather initial info → correlate source → deep investigation → session management) are clearly sequenced, with Step 3 marked CRITICAL and an explicit worked correlation example. There are no explicit validation checkpoints after session initialization (e.g. verifying the session is alive or that symbols loaded), fitting 'clear sequence with most checkpoints present; minor validation gaps'; the destructive/batch cap does not apply since this is read-only investigation. | 4 / 5 |
Progressive Disclosure | The skill is ~215 lines with no bundle files at all: everything — Common Debugging Patterns, Examples, Related Skills, Links — is inlined in SKILL.md. Section organization is good, but content like the debugging patterns and examples is a natural fit for one-level-deep reference files, matching 'some structure but could be better organized; content that should be separate is inline'. The under-50-line exception does not apply. | 3 / 5 |
Total | 15 / 20 Passed |