v2.1.0

2.1.0 (2026-08-30)

Features

• spec & docs & review redux — contract-forward docs, adversarial review, loop skill

• feat(skills): rebuild spec and review pipeline with validation gates

Store

• add shared store-init script with unit tests, vendored into nine
  skills through the keyed vendor-assets manifest
• replace inline bash copies and wire store init into the explore,
  pr, docs, and qa workflows that touch .codevoyant

Spec

• tabulate enumerable sets with completeness gates: tabulation.md
  contract, tables/ directory, SCOPE=tabulation validation
• refuse uncalled-out cross-module writes in doc-aware mode via
  boundary callouts, [BLOCKED] entries, and review audits
• add constrained Requirements block with SCOPE=requirements
  validation in the plan template and spec review

Review

• add adversarial review dimension: red-team-adversary (Dimension 5)
  and claim-checker agents
• add security gates, CI/commit/static-floor checks; restrict to
  file-level comments with a fixed minimal publish fallback

Docs

• contract-forward prose-minimal policy: closed LLM text budget and
  @Human markers across twelve templates
• CLI artifact validation gate with pinned mermaid renderer and
  semantic caps; refresh pr, docs, and spec pages and add loop page

Requirements

• R1-R7 constrained requirements authoring with gate semantics,
  enforced in docs review and spec validation

Loop

• new bounded repeat-until-objective loop skill with runner and
  judge background agents and a max-iterations bound

Housekeeping

• replace CLAUDE.md with AGENTS.md, drop Claude Code config, and
  move the changelog sanitizer to a mise task

• fix(vendor): respect the files filter so shared tests stay in shared

• vendor-assets treated files as a union with the whole source dir;
  it is an exclusive filter per the documented contract
• re-vendor: drop the 10 vendored test_cv_init_store.py copies; the
  test lives only in skills/shared/store-init/ (still run by mise test)
• update the vendor-assets tests to the new contract (unlisted files
  stay put; assets without files still walk the whole source)

• refactor(loop): collapse to a single-command run with a tracking doc

• drop the flow-style definition model: no new/go/list/status workflows,
  no saved loop definitions — loops are not flows
• /loop --until [--max N] [--check ] [--resume ]
  writes the tracking doc (.codevoyant/loops/{slug}/loop.md) and runs
  immediately; the procedure lives inline in SKILL.md
• merge loop-runner + loop-judge into one loop-agent that performs the
  task and strictly judges the objective per iteration
• update docs/skills/loop.md to match