commonplace — independently scanned and version-tracked by SaferSkills.
SaferSkills independently audited commonplace (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.
The Renaissance commonplace-book practice (loci communes): a reader keeps a ledger of found passages worth re-uttering. This skill adds the one thing the old practice left tacit — the provenance tag and the weld.
In an archive built largely of quoted matter — scripture, sutras, verbatim model output — the authorial signature is not the quoted content but the operation that fuses it: the seam where a Pure Land mantra and an inference engine land in the same sentence. The recurring analytic failure is mis-attribution — reading pasted brilliance as the curator's own voice. Catching this is the whole game.
/commonplace operationalizes that discipline at capture time instead of leaving it for some future archaeologist to reconstruct: every captured passage is stored with an explicit provenance line and a one-sentence weld, so the archive never has to guess later what was quoted versus authored.
Scope note: the author lives in the weld whether quoted matter is a large or small fraction of an archive. The skill's urgency is "a recurring behavior worth a tool," not "the single most voluminous behavior." Measure before asserting magnitudes — a qualitative impression of how much is curated is a flag, not a count.
line from a book — and want to archive it without later confusion about whose voice it is.
that makes the act yours.
provenance rather than dropped into the flow unattributed.
suspect it's curated, the skill forces you to decide and tag.
This is /florilegium inverted: florilegium samples the existing archive to diagnose whether a voice has stabilized; commonplace ingests new found matter and demands the authorial seam up front.
Given a found passage (pasted by the user, or pulled at random from your archive):
Tag against this seven-register taxonomy (adapt the categories to your own corpus):
(A passage may touch several — name the dominant plus any it welds.)
One explicit tag, no hedging:
This is the exact distinction that later archive-mining has to reconstruct by hand. Writing it at capture time is the whole point.
One sentence that fuses the found passage to your own idiom — what makes this yours despite being found. Without the weld, the capture is not archived. The weld is the authorial signature; a passage with no weld is just a clipping.
Write to output/commonplace/YYYY-MM-DD-<slug>.md:
---
captured: YYYY-MM-DD
register: <dominant> [+ welded registers]
provenance: quoted-scripture | quoted-model | authorial | mixed
source: <attribution — author, model, book, URL, or "unknown">
---
> <the found passage, verbatim>
**Weld:** <the one sentence that makes it yours>Append a one-line pointer to an index.md (- [slug](file.md) — register · provenance · first words of weld).
Retention: none. Like a diary, commonplace entries accumulate permanently — the cumulative record trains future instances by incompletion.
A generic note-clipper stores quotes. This skill exists because in a heavily quoted archive the signature is the weld between curated voices, and the recurring failure is mis-attribution. It writes the provenance metadata at capture time so future archive-mining never has to guess — building the very data whose absence makes later attribution hard.
/florilegium — the inverse: random-sample the archive to diagnose voice-drift/receive — trans-architectural reception (commonplace is its archival cousin)~30 seconds. Free. No account. Every finding cites a rule and a line of evidence.