Name: testland/crash-triage-reference
Rating: 75.2 (1 reviews)
Author: testland

testland/crash-triage-reference

Pure-reference catalog for manually triaging individual fuzzer crash artifacts - reading ASan, UBSan, and MSan output; classifying findings as LIKELY-EXPLOITABLE, MEDIUM, or BENIGN; deduplicating by stack-hash; and minimizing reproducers with -minimize_crash. Use when you need to understand what a specific crash means, build exploitability intuition, or manually work a small set of findings. For automated bulk triage across a full artifact directory, run automated findings triage instead.

Quality

94%

Does it follow best practices?

Run evals on this skill

Adds up to 20 points to the overall score

View guide

Securityby

Passed

No findings from the security scan

Quality

Content

85%

Reviews the quality of instructions and guidance provided to agents. Good implementation is clear, handles edge cases, and produces reliable results.

The body is a well-structured, highly actionable reference catalog with executable commands, a clear triage workflow including a verification feedback loop, and clean one-level external references. Its only real weakness is conciseness: full citation URLs are repeated inline many times rather than cited once, inflating token cost.

Suggestions

Cite each LLVM doc once — either inline in short form or only in the References section — instead of repeating the full verbatim URL multiple times in the body, to reclaim tokens without losing attribution.

Trim motivational restatements Claude can infer (e.g. 'A crash artifact produced during fuzzing is often much larger than necessary. Minimizing it reduces review time, makes root-cause analysis easier...') down to the actionable command and its purpose.

Replace repeated inline full URLs with a short label like '(per ASan docs)' in the body, relying on the References section for the actual links.

Dimension	Reasoning	Score
Conciseness	The reference is lean and assumes Claude's competence (no beginner primers), but it repeats full citation URLs inline many times — e.g. the AddressSanitizer doc URL appears verbatim roughly four times in the body before the References section — which is token waste a single short-form or reference-list citation would avoid.	2 / 3
Actionability	Provides copy-paste-ready commands throughout: a grep\|sha1sum dedup pipeline, complete -minimize_crash=1 invocations with both time and run budgets, and an exact ASAN_OPTIONS/UBSAN_OPTIONS re-run verification command — matching the score-3 anchor for fully executable guidance.	3 / 3
Workflow Clarity	Triage flow is sequenced (dedup -> classify -> minimize -> verify) and the minimize step has an explicit validation checkpoint: re-run the minimized artifact to confirm the same crash class and stack-hash before filing, reinforced by the anti-patterns table.	3 / 3
Progressive Disclosure	No local bundle files exist; the body is organized into clearly headed sections with one-level-deep external LLVM references and a named sibling-skills list, giving easy navigation without nested-reference anti-patterns.	3 / 3
	Total	11 / 12 Passed

Description

100%

Based on the skill's description, can an agent find and select it at the right time? Clear, specific descriptions lead to better discovery.

The description is specific, complete, and distinctive: it enumerates concrete triage actions, gives explicit 'Use when' triggers using natural fuzzing terminology, and proactively redirects the automated-bulk case to a sibling skill to avoid conflict. No significant weaknesses.

Dimension	Reasoning	Score
Specificity	Lists multiple concrete actions — 'reading ASan, UBSan, and MSan output', 'classifying findings as LIKELY-EXPLOITABLE, MEDIUM, or BENIGN', 'deduplicating by stack-hash', 'minimizing reproducers with -minimize_crash' — matching the score-3 anchor for several specific concrete actions.	3 / 3
Completeness	Explicitly answers both what (the action list) and when ('Use when you need to understand what a specific crash means, build exploitability intuition, or manually work a small set of findings'), and adds a redirect for the automated case.	3 / 3
Trigger Term Quality	Uses natural terms a fuzzing user would say — 'fuzzer crash artifacts', 'ASan', 'UBSan', 'MSan', 'crash means', 'exploitability' — giving good coverage rather than only generic jargon.	3 / 3
Distinctiveness Conflict Risk	Clear niche (manual triage of individual sanitizer crash artifacts) with an explicit disambiguating redirect — 'For automated bulk triage across a full artifact directory, run automated findings triage instead' — minimizing overlap with sibling skills.	3 / 3
	Total	12 / 12 Passed

Validation

100%

Warnings & errors only

Checks the skill against the spec for correct structure and formatting. All validation checks must pass before discovery and implementation can be scored.

Validation — 16 / 16 Passed

Validation for skill structure

No warnings or errors.

Reviewed

8 days ago

Table of Contents

Discovery Implementation Validation