Rubric index
This page is the index over every normative rubric document muster
publishes under docs/rubric/. It exists to satisfy the charter’s
cite-a-source rule (“every check cites a normative source, upstream or
ours”) mechanically rather than aspirationally: a rubric a checker cites
but that isn’t listed here, or a citation inside a listed rubric that has
drifted off its real target, is a defect this index and its companion
script (scripts/check-rubric-citations.mjs) are built to surface.
citation-count below is informational only — it counts how many
`symbol` (`file:line[-line]`) code anchors a rubric file currently
makes. drift-check-status is populated by the script, not
hand-maintained — do not hand-edit either column expecting it to stay
accurate; run node scripts/check-rubric-citations.mjs to get the real,
current answer. This index’s own text only records which adapter(s)/check(s)
cite each document, which is the one column the script does not (and
cannot) infer on its own.
| rubric-file | cited-by | citation-count (informational) | drift-check-status |
|---|---|---|---|
crosslayer-contradiction-gate.md | src/crosslayer/contradiction-lint.ts (the cross-layer contradiction lint’s subject-matter gate; findings carry the in-code constant MUSTER_RUBRIC_CITATION, not yet a file:line anchor to this document) | 0 | populated by script |
memory-utilization-taxonomy.md | src/adapters/memory-utilization/ (the memory-utilization adapter’s rubricCitation field on every emitted finding) | 0 | populated by script |
skills-trigger-taxonomy.md | src/adapters/skills/trigger.ts, src/adapters/skills/types.ts (the Agent Skills adapter’s behavioral trigger-conformance grader, runTriggerConformance) | 15 | populated by script |
sop-rule-taxonomy.md | src/adapters/openclaw-sop/ (every SOP manifest entry’s source.normative field cites this document by path, per FR-009 of the SOP adapter’s own spec) | 0 | populated by script |
spec-kitty-behavioral-axes.md | Not consumed by any muster adapter. Written to unblock the downstream M4 mission (Spec Kitty issue MOES-Media/spec-kitty#24), which pastes this document’s rubricText blocks verbatim into its own JudgeAssertions — muster’s own spec-kitty-profile adapter is entirely static and does not use it. Its Integration Contract section cites muster’s own SOP judge (src/adapters/openclaw-sop/judge.ts) as prior art for one structural detail, in a comma-separated style the checker doesn’t parse (see “Known blind spot” below). | 0 | populated by script |
spec-kitty-profile-taxonomy.md | src/adapters/spec-kitty-profile/ (the Spec Kitty agent-profile static conformance adapter) | 0 | populated by script |
These six documents live under docs/rubric/ in the repository; this
Starlight page mirrors docs/rubric/index.md in substance and does not
duplicate them as separate site pages.
Known blind spot
Section titled “Known blind spot”scripts/check-rubric-citations.mjs only recognizes the canonical
`symbolName` (`file:line[-line]`) grammar — the symbol and the
(`file:line`) group must be adjacent, both backtick-quoted. Two
pre-existing citations in spec-kitty-behavioral-axes.md predate this index
and are invisible to the checker by design, not silently passed as
clean, simply never checked: a fully bare “(lines 62-67 of that file)”
reference (no file path or backticks at all), and a comma-separated
`buildJudgeSystemPrompt`, `src/adapters/openclaw-sop/judge.ts:62-67`
mention (both halves backtick-quoted, but joined by a comma rather than the
canonical symbol (`file:line`) adjacency). No FR licenses editing that
document’s prose, so both are left as-is; see the recorded-gaps
register if a future change wants to close
this gap.
Normative sources
Section titled “Normative sources”This index itself makes no grading claims — it only points at the six
documents that do. See each document’s own “Normative Citation” (or
equivalent) section for its upstream/[MUSTER-OWN] provenance.