docs: update docs/architecture/workflows.md to document missing eval npm commands

[github-actions]

Component

Documentation

Bug Description

docs/architecture/workflows.md was updated in commit 18ce4c6 (PR #1834 — feat(agents): add Vally evaluation agents and prompts), but the npm command reference table is missing approximately 20 eval-related commands that exist in package.json. The documented eval commands cover only the lint and basic run variants:

Documented (10 eval entries):
eval:lint:vally, eval:lint:schema, eval:lint:text, eval:lint:safety, eval:lint:skills, eval:run, eval:run:skills, eval:run:agents, eval:run:scripts, eval:compare

Missing from the reference table:

Command	Purpose
eval:presence	Checks that every changed AI artifact has an eval-spec backlink
eval:execute	Runs vally evals for each changed AI artifact per the manifest
eval:moderate	Runs content moderation on a single artifact
eval:moderate:corpus	Runs corpus-wide content-moderation evaluation
eval:moderate:artifacts	Runs artifact-level moderation per manifest
eval:moderate:test	Runs Pester tests for the moderation runner
eval:run:equivalence	Runs the baseline-equivalence eval suite
eval:behavior-prompts	Runs behavior-conformance evals for prompts
eval:behavior-instructions	Runs behavior-conformance evals for instructions
eval:behavior-skills	Runs behavior-conformance evals for skills
eval:equivalence	Runs the baseline equivalence comparison
eval:dashboard	Renders equivalence results into a dashboard
eval:agent	Runs the agent-matrix evaluation
eval:agent:matrix	Runs the full nightly agent-matrix suite
eval:agent:matrix:dryrun	Dry-run mode for the nightly agent-matrix suite
eval:agent:changed	Runs agent-matrix only for changed AI artifacts
eval:agent:dashboard	Renders agent-matrix results into a dashboard
eval:agent:dashboard:open	Renders and opens the agent-matrix dashboard
eval:agent:report	Full agent-matrix run and dashboard open
eval:agent:report:dryrun	Dry-run agent-matrix run and dashboard open

Expected Behavior

docs/architecture/workflows.md should include a complete row for each eval-related npm command so contributors can discover and understand the full evaluation toolchain from the reference table.

Steps to Reproduce

1. Open docs/architecture/workflows.md.
2. Locate the npm command reference table.
3. Compare against package.json — eval:presence, eval:execute, all eval:moderate:*, all eval:agent:*, all eval:behavior-*, and eval:run:equivalence are absent.

Additional Context

• Commit introducing the eval agent-matrix commands: 18ce4c6 (PR feat(agents): add Vally evaluation agents and prompts #1834).
• The eval commands are used directly by CI jobs in .github/workflows/pr-validation.yml and .github/workflows/eval-validation.yml.
• Related open issue: docs: update scripts/README.md to document the new evals orchestration scripts #2012 tracks the scripts/README.md gap for evals orchestration scripts.
• docs/contributing/evals-ci.md documents the eval CI contract but does not serve as a full npm command reference.

Generated by Documentation Update Check · 1K AIC · ⌖ 22.3 AIC · ⊞ 34.4K · ◷