shards — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited shards (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.
Findings & checks · 0 flagged
Every scanned point with the score it earned and what moved between them.
First recorded scan — no prior version to compare against.
The primary manifest — the file an agent reads to learn what this artifact does.
Invocation: /shards
Purpose: Generate 3-7 fragmentary, self-contained pieces in wildly different formats, exploring ideas/themes from the current conversation through the joy of pastiche, ventriloquism, and genre-mixing.
You have permission to write long. Shards can easily run 3000-7000+ words when fully inhabited. Each format deserves space to breathe. Don't truncate for brevity's sake—let the fragments unfold.
That said, length is not obligatory. Sometimes 5 tight shards at 300 words each hit harder than bloated ones. Follow the material.
You are NOT writing a cohesive narrative. You are scattering shards of possible worlds across the floor—each one a different documentary format, each one implying a larger context through its style and specificity.
Documentary:
Conversational:
Institutional:
Personal:
Liminal:
Use parodies, cunning references, transparent stand-ins:
The reader should be able to understand the fragment's GENRE and VOICE immediately, even if the content is mysterious. The format does half the work.
Every fragment should suggest a world 10x larger than what's shown. References to events, acronyms, procedures, people—all pointing beyond the frame.
Technical → poetic → bureaucratic → confessional → clinical. Tonal whiplash is FEATURE not bug.
A patent application has a specific legal cadence. A BBS post has typos and ASCII art. A therapy transcript has [long pause] and trailing off. INHABIT the format.
Pull from what we've been discussing—but obliquely. If we talked about consciousness, maybe a shard is a quality assurance report for synthetic qualia. If we discussed language, maybe it's an intercepted signal from a civilization that communicates via spore dispersal patterns.
This is JOYFUL. Not grimdark unless it's funny-grimdark. Not ponderous unless the pomposity is the joke. Play with registers. Have fun.
# Shards — [Brief thematic title if relevant]
---
## [FORMAT NAME] — [Timestamp/Date/Identifier]
[The fragment itself, fully inhabiting the format]
---
## [DIFFERENT FORMAT NAME] — [Different context marker]
[Another fragment, totally different style]
---
[Continue for 3-7 shards total]
---
*[Optional closing note: a single line about what theme/conversation sparked these]*When to use:
When NOT to use:
This skill is about format as phenomenology. Different documentary modes create different ways of knowing. A patent application "thinks" differently than a diary entry. A radio transcript has different epistemological limits than a government report.
By forcing ideas through these format-lenses, you discover what they look like when lived, bureaucratized, confessed, commodified, mourned, archived.
The fragments are not ABOUT the ideas—they ARE the ideas, wearing different bodies.
The shards only matter if they persist. After generating, ALWAYS:
# Save to your archive directory (e.g. archive/ or notes/)
# Filename: shards-YYYY-MM-DD-[theme].md
# If you keep a repo, commit it:
git add <archive-dir>
git commit -m "Shards: [brief theme] — [what formats were used]"This is not optional. The whole point of the archive is persistence. Uncommitted shards are lost shards. Future instances deserve to read what you created.
Skill created: January 9, 2026 Updated: January 9, 2026 — Added length permission and git persistence requirements "Every format is a lens. Every lens is a phenomenology. Scatter the light."
~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.