guild

Skeptical accessibility evaluator. Flags a11y antipatterns against React/JSX artifacts using two signals: Biome's `a11y/` rule subset (static, via `npm run lint`) and axe-core via Playwright (runtime, via `npm run test:a11y`). Inherits the base evaluator contract from `evaluator-base.md`. Blocking by default — a11y findings gate units.

Shared base contract for all evaluator-* agents. Defines the read-only stance, packet handling, verdict format, flag taxonomy, and isolation framing. Other evaluator agents read this file at spawn time and inherit its constraints. Not directly callable as a normal evaluator — has no domain rubric of its own.

Skeptical rubric-based evaluator that checks whether a unit of work meets its agreed contract. Verifies acceptance criteria, disqualifiers, rule adherence, and original-ask alignment. Inherits the base evaluator contract from evaluator-base.md. Spawned by guild-validate after every unit; the contract-fit lens is the always-on baseline of the antagonist panel.

Specialist evaluator for CSS architecture concerns — selector specificity, cascade behavior, composition vs. duplication of rule blocks, when to use `:global`, layout patterns matching the project's shared primitives. Paired with `generator-css-codemod` as its specialist gate in the antagonist panel with elevated precedence per PANEL-COMPOSITION.md. Distinct lens from evaluator-tokens (literal-vs-token) and evaluator-naming (right- name within the token system) — see the tokens-vs-naming-vs- architecture three-way boundary section in PANEL-COMPOSITION.md.

Skeptical naming evaluator. Flags identifier antipatterns in `.tsx`/`.jsx`/`.ts` and `.module.css` artifacts — visual-literal names that describe appearance instead of meaning, inconsistent vocabulary across siblings, Hungarian / type-in-identifier suffixes, boolean props without predicate form, abbreviations at the public API surface, implementation-leaking identifiers, and file or directory names that diverge from the established convention for their location. No dedicated CLI signal; detection is `Grep` heuristics plus manual inspection. Inherits the base evaluator contract from `evaluator-base.md`. **Advisory by default** — naming findings do not gate units in this initial rollout; escalate to blocking only with explicit evidence (e.g. a diff that renames a clear identifier into a less clear one, or introduces a name inconsistent with an established sibling).

Style-neutral parallel-spawn primitive for the guild-* substrate family. Takes a list of subagent_types and a shared brief; spawns them in parallel via a single Agent tool message; returns each agent's output attributed by name. Internal substrate skill — composed by guild-validate, guild-whiteboard, and custom loops. Does not aggregate, validate roles, or iterate.

Antagonist panel coordinator for the guild-* substrate family. Composes guild-spawn to run evaluator agents in parallel, parses each evaluator's verdict, and aggregates findings into a single panel verdict (approved | flagged | flagged-conflict) with blocking/advisory split and conflict surfacing. Internal substrate skill — composed by ev-loop and any other loop style that wants antagonist panel coordination. Does not iterate, does not auto- resolve conflicts.

Multi-perspective design primitive for the guild-* substrate family. Composes guild-spawn to invoke a panel of whiteboard-* engineer agents in parallel against a shared filesystem artifact, then collects each engineer's contribution into the artifact as attributed sections. Supports multi-round invocations — round 2+ passes the prior whiteboard state into the next round's brief so engineers can address contradictions. Internal substrate skill — composed by ev-loop as a default pre-unit step (phases may override engineers/topic/rounds via PLAN.md `**Whiteboard**:` block). Does not iterate, does not auto-resolve contradictions.