Add Claude Code CLI as a ringer engine

README.md

@@ -397,6 +397,7 @@ Every community PR that lands in main is credited here — that's a project rule
 - [@davekopecek](https://github.com/davekopecek) (Dave Kopecek) — committed the design-reference fixture so the design-token guard runs on every machine (#30)
 - [@snapsynapse](https://github.com/snapsynapse) (Sam Rogers) — graceful shutdown on SIGINT/SIGTERM with worker-tree cleanup and finished state, plus the 14-test end-to-end CLI regression suite (#4)
 - [@mlava](https://github.com/mlava) (Mark Lavercombe) — named setup failures across every diagnostic surface (#37), `run --baseline`, the no-workers check preflight (#38), guidance on check-writing failure modes (#57), early warnings for missing worker commands (#59), and preserving fix-swarm patches across retries (#56)
+- [@hcoronel1](https://github.com/hcoronel1) (Hernán Coronel) — added Claude Code CLI as a ringer engine, with a `claude-code-json` result parser that keeps API/infra failures out of the scoreboard without hiding real agent failures
 
 Contributions are welcome — see [CONTRIBUTING.md](CONTRIBUTING.md) for the philosophy and what gets a PR merged fast. The short version: small and scoped, rebased on current main, every claim backed by an executed test. Authorship is always preserved — where a maintainer pushes a mechanical fix to your branch, you remain the commit author.
 

config.sample.toml

@@ -91,6 +91,19 @@ token_regex = "tokens\\s+used\\s*:?\\s*([0-9][0-9,]*)"
 # default reads the `model:` header; override it here only if that format changes.
 model_report_regex = "(?m)^model:[ \\t]*([^ \\t\\r\\n]+)[ \\t]*\\r?$"
 
+# Claude Code CLI (opt in). Claude Code permissions are not an OS filesystem
+# sandbox; use an OS-level wrapper when filesystem containment is required.
+# [engines.claude]
+# bin = "claude"
+# model_default = "claude-sonnet-5"
+# args_template = [
+#   "-p", "--model", "{model}", "--output-format", "json",
+#   "{access_args}", "{engine_args}", "{spec}",
+# ]
+# sandbox_args = ["--permission-mode", "acceptEdits"]
+# full_access_args = ["--dangerously-skip-permissions"]
+# result_parser = "claude-code-json"
+
 # Grok Build CLI (xAI). Install: curl -fsSL https://x.ai/cli/install.sh | bash
 # (or: npm install -g @xai-official/grok), then `grok login` — OAuth on a
 # SuperGrok or X Premium Plus plan. Headless mode (-p) runs a full agentic

registry/model-identity.toml

@@ -71,6 +71,68 @@ access = "OpenRouter API"
 # names. Unlisted slugs are marked unregistered and derive a display name;
 # their complete raw value appears only in scoreboard diagnostics until verified.
 
+[engines.claude]
+harness = "Claude Code CLI"
+access = "Claude subscription"
+default_model_key = "claude-sonnet-5"
+
+[engines.claude.models."claude-sonnet-5"]
+display = "Claude Sonnet 5"
+lab = "Anthropic"
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude.models."sonnet"]
+display = "Claude Sonnet 5 (alias)"
+lab = "Anthropic"
+alias = true
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude.models."claude-opus-5"]
+display = "Claude Opus 5"
+lab = "Anthropic"
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude.models."claude-fable-5-1"]
+display = "Claude Fable 5.1"
+lab = "Anthropic"
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude.models."claude-haiku-4-5"]
+display = "Claude Haiku 4.5"
+lab = "Anthropic"
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude-sonnet]
+# Legacy engine name retained for historical rows; the harness is unchanged.
+harness = "Claude Code CLI"
+access = "Claude subscription"
+default_model_key = "claude-sonnet-5"
+
+[engines.claude-sonnet.models."claude-sonnet-5"]
+display = "Claude Sonnet 5"
+lab = "Anthropic"
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
+[engines.claude-sonnet.models."sonnet"]
+display = "Claude Sonnet 5 (alias)"
+lab = "Anthropic"
+alias = true
+confidence = "verified"
+source = "https://docs.anthropic.com/en/docs/about-claude/models/overview"
+last_verified = 2026-09-16
+
 [engines.opencode.models."openrouter/z-ai/glm-5.2"]
 display = "GLM 5.2"
 lab = "Z.ai (Zhipu AI)"