Stats
Actions
Tags
From shatter
Write a markdown file that enumerates issues found during a Shatter review and includes relevant system and project context. Use when a downstream user wants a durable issue report instead of tracker-specific automation.
How this skill is triggered — by the user, by Claude, or both
Slash command
/shatter:report-shatter-issuesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Recommended model: mid. Structured output assembly with light judgment on
Recommended model: mid. Structured output assembly with light judgment on which observations rise to the level of an issue.
Create a markdown report file, not a tracker ticket.
The report should be durable, portable, and detailed enough that a user can keep it locally or paste it into GitHub later.
run-shatterBefore writing the report, collect environment and project context with:
scripts/collect-context.sh --run-dir <run-dir> --target <path> --artifact <path> ...
Save that output alongside the report and include it in the final markdown file.
The report must be a markdown file with:
Run summaryEnvironment and project contextEnumerated issuesEvidence and artifact referencesFor the exact issue schema, read
../../run-shatter/references/report-schema.md.
Include only actual issues or clearly labeled uncertainties. Do not restate every observed behavior.
Good issue categories:
If the review found no issues, still write the markdown file and state that no actionable issues were found.
For each issue, include:
Prefer one numbered issue section per finding.
scripts/collect-context.sh (bundled with this skill)npx claudepluginhub shatterproof-ai/agents --plugin shatterCreates, edits, and optimizes skills for Claude Code, including drafting, evaluating with test prompts, iterating on performance, and improving skill descriptions for better triggering accuracy.