Element Types¶

FORMAT · ELEMENT TYPES

Every .md file in the model tree is one element. The type: field in YAML frontmatter selects the element type. A value outside the type inventory below is stored as Unknown and raises error E005 (unrecognised type:); a missing type: is E001. The authoritative list is ElementType::ALL in crates/syscribe-model/src/element.rs — every type there appears on this page.

Definitions¶

Usages¶

Requirement is listed with the native elements below (the native handler owns the type).

Native elements (own schema)¶

These are not standard SysML usages — they carry a stable opaque identifier and their own required field sets. They are id-identified: their identity is the stable id, and their human-readable label is name (free prose — spaces and punctuation allowed, W042 does not apply). The title field is removed; a title: on any element is error E025, and E024 (formerly: name on an id-identified type) is retired. See Frontmatter → Label field.

Baseline (ADR-SYS-BASELINE-001) is a sealed, commit-anchored release snapshot of a model scope (frozenScope:), created by syscribe baseline create and re-checked on every validate: drift is E520 when released, W520 when approved, silent for draft, skipped for superseded; seal/manifest tamper is E521, an unresolved supersedes: E522. Commands baseline create/verify/diff/list/show.

FeatureModel authors a whole feature tree in one file: each featureTree: entry's name: is a dot-separated path relative to the sheet (Platform.CortexM), id: is optional (derived as FEAT-PLATFORM-CORTEXM when omitted), and optional crossTreeConstraints: / parameterConstraints: sit on the same sheet. Codes E231–E233, W048. See Variability.

Zone / Conduit (§13) model IEC 62443 industrial cybersecurity: a Zone (ZN-*) groups parts under a Security Level (targetSL/achievedSL, each 1–4, else E925); a Conduit (CD-*) connects two zones. Both take status: draft/review/approved/deprecated (else E926). Structural elements may carry targetSL/achievedSL/inZone:. Validation E950–E956, E925, E926, W950–W953; commands zones, conduits, zones --coverage.

ReviewRecord (§19) captures a formal review event (design / requirements / hazard / test-readiness review, inspection, walkthrough) and the model elements it covers — a thin, baselined traceability anchor whose recordedAt: points to the external review (e.g. a GitHub PR). Validation E700–E705, W700, W704; commands reviews, review, reviews --coverage. See CLI → Reviews.

TradeStudy (§15) records a weighted-criteria evaluation of design alternatives (criteria with weight + maximize/minimize, alternatives, a scores matrix, optional objective/decision). The tool computes — never writes — normalised/weighted scores and rankings. Validation E869–E877, W061–W064; command trade-study.

PlanningItem (ADR-SYS-PLANITEM-001) is the model's native representation of the day-to-day work of getting from Requirement to satisfied/verified — the shape a Jira epic/story/task or a GitHub issue hierarchy fills today, made durable and structurally part of the traceability graph. A strict single-parent tree (parent:, at most one — not a DAG); a top-level item (no parent:) must set achieves: (one or more Requirements (or a goal, ADR, argument, plan or baseline) this branch of work exists to realise, deliberately a separate field from satisfies:, which stays scoped to architecture semantics). status (todo/in_progress/blocked/done) and itemType (bug/task/feature) reuse GitHub's own current vocabulary verbatim. blockedBy: names one or more elements it's waiting on — resolved permissively like evidence.ref: below, not restricted to PlanningItem — with dangling and cycle checks; a non-empty blockedBy: while status isn't blocked warns (likely stale), but status: blocked with an empty blockedBy: raises nothing — being blocked needs no proof, unlike claiming done. evidence: is a list of duck-typed entries — ref: (any resolvable element, unrestricted by kind) or path: (a file/doc, resolved like implementedBy:) — each with an optional rationale: that waives that one entry's own check. A leaf item (no children) claiming status: done must have at least one non-waived, resolving evidence: entry — graded harder than the analogous Requirement rule (W300, a warning) since claiming done with no proof is a correctness defect, not a time-bound gap. assignedTo: names a single Unix-style username — not a cross-reference (users aren't model elements) — always format-checked (^[a-z_][a-z0-9_-]{0,31}$, E723), and additionally checked against a project-declared roster ([users] in .syscribe.toml, mapping username → display name) only when that roster is non-empty (E722); roster membership is dormant otherwise, matching every other opt-in .syscribe.toml-configured table — a malformed roster key is W309 and excluded from the effective roster. show resolves and prints the declared display name alongside the username. A done item whose achieves: Requirement lacks the verification bar W002/W305 apply (an active TestCase for a leaf, an active L3–L5 TestCase for a parent) warns W310; two active items (in_progress, or carrying claimedBy:) that share an achieves: Requirement or an evidence[].path warn W311 (possible duplicate work). For concurrent (multi-agent) work, syscribe claim <PI-id> --by <agent-id> sets the advisory claimedBy:/claimedAt: markers (refused on a done item or one claimed by someone else) and syscribe release <PI-id> clears them. Everything else is queried via the generic list/show/ls/find/refs commands and written via set or the guarded MCP create_element/update_element/etc. tools. Validation E706–E717, E719–E723, W308–W311. See examples/planning-item/ for a complete worked example.

Tier 2 — Safety & cybersecurity elements (own schema)¶

These types support ISO 26262 HARA and ISO/SAE 21434 TARA workflows. Each carries a stable opaque identifier and validated required fields.

All of these require id, name and status. Every stable id needs at least one 2–12-character category segment before the numeric suffix — DS-BRAKE-001, not DS-001.

Tier 4 — Safety analysis containers¶

These analysis types use one of two authoring patterns:

• Option A (file-per-element) — each node is its own .md file; the parser loads them individually.
• Option B (exploded container) — a single container file holds all rows in frontmatter tables; the walker synthesises a first-class element per row so all cross-reference and query machinery works without changes.

TARASheet explodes each row into the appropriate Tier 2 type (DamageScenario, ThreatScenario, CybersecurityGoal, SecurityControl) at parse time.

See Safety Analysis for authoring examples.

Namespace elements¶

Diagram elements¶

See Diagrams for the full diagramKind list.