red-queen

{"kind":"meta","skill":"red-queen","version":"8.0.0","updated":"2026-02","format":"jsonl-progressive"}
{"kind":"liza","path":"~/.claude/skills/red-queen/liza-advanced.nu","usage":"L=\"$HOME/.claude/skills/red-queen/liza-advanced.nu\"; nu $L <cmd>"}
{"kind":"principle","id":"deterministic_over_ai","text":"If it can be computed, compute it. AI generates test commands only.","ai_does":["Generate test commands","Fix code","Read docs","Discover promises"],"code_does":["Execute tests","Select survivors","Lock regressions","Gate transitions","Score landscape","Validate lineage"]}
{"kind":"principle","id":"exit_codes_truth","text":"Exit codes are ground truth, not AI interpretation of output.","survivor_check":"exit_code != expect_exit","no_ai_judgment":"AI never decides if something passed - shell exit code decides"}
{"kind":"principle","id":"done_when_lineage","text":"Every survivor is a permanent shell command with expected exit code.","storage":"done_when in blackboard YAML","ratchet":"validate runs ALL done_when, deterministic pass/fail"}
{"kind":"principle","id":"file_beads_immediately","text":"br create happens same moment as task-add-check. Never defer.","timing":"File bead at survivor selection, not at session end"}
{"kind":"state_machine","id":"transitions","flow":"UNCLAIMED → claim → IN_PROGRESS → coder-submit → READY_FOR_REVIEW → validate → APPROVED → merge → MERGED"}
{"kind":"gate","id":"assert_can_submit","check":"Status is IN_PROGRESS, agent_id set","code":"liza lines 132-139"}
{"kind":"gate","id":"assert_can_review","check":"Status is READY_FOR_REVIEW, commit SHA present, validation results exist","code":"liza lines 142-156"}
{"kind":"gate","id":"assert_can_merge","check":"Status is APPROVED, review decision is APPROVED","code":"liza lines 159-167"}
{"kind":"gate","id":"assert_no_test_weakening","check":"Changed files don't touch tests/ unless explicitly allowed","code":"liza lines 170-177"}
{"kind":"cmd","id":"init","usage":"nu $L init","desc":"Initialize state machine"}
{"kind":"cmd","id":"task_add","usage":"nu $L task-add <name> --spec_ref README.md","desc":"Create task with spec reference"}
{"kind":"cmd","id":"task_add_check","usage":"nu $L task-add-check <task> \"<cmd>\" --expect_exit=0","desc":"Register done_when check"}
{"kind":"cmd","id":"claim","usage":"nu $L claim <task> red-queen","desc":"Claim the session"}
{"kind":"cmd","id":"gen_start","usage":"nu $L gen-start <task>","desc":"Start generation N"}
{"kind":"cmd","id":"gen_survivor","usage":"nu $L gen-survivor <task> \"<dim>\" \"<cmd>\" --severity <SEV>","desc":"Record survivor + file bead + update landscape"}
{"kind":"cmd","id":"gen_discard","usage":"nu $L gen-discard <task> \"<dim>\"","desc":"Record discard (no bug)"}
{"kind":"cmd","id":"landscape","usage":"nu $L landscape <task>","desc":"Compute fitness from blackboard"}
{"kind":"cmd","id":"coder_submit","usage":"nu $L coder-submit <task> red-queen","desc":"Transition to READY_FOR_REVIEW"}
{"kind":"cmd","id":"validate","usage":"nu $L validate <task>","desc":"Run ALL done_when checks"}
{"kind":"cmd","id":"lineage_replay","usage":"nu $L lineage-replay <task>","desc":"Verify current code defeats entire lineage"}
{"kind":"cmd","id":"carnage","usage":"nu $L carnage <task>","desc":"Track kill rate and lethality"}
{"kind":"cmd","id":"escalate","usage":"nu $L escalate <task>","desc":"Auto-promote severity for bleeding dimensions"}
{"kind":"cmd","id":"show","usage":"nu $L show --task=<task>","desc":"Display blackboard state"}
{"kind":"cmd","id":"mutate","usage":"nu $L mutate <task> <dir> --file <f> --function <fn>","desc":"Run cargo-mutants, missed → survivors","ref":"references/automated-weapons.md"}
{"kind":"cmd","id":"spec_mine","usage":"nu $L spec-mine <task> <dir> --bin <bin>","desc":"Extract promises from README/CLI/doctests","ref":"references/automated-weapons.md"}
{"kind":"cmd","id":"quality_gate","usage":"nu $L quality-gate <task> <dir>","desc":"Run FP/DRY/coverage checks","ref":"references/automated-weapons.md"}
{"kind":"cmd","id":"fowler_review","usage":"nu $L fowler-review <task> <dir> --complexity-threshold 15","desc":"AST analysis + test smells + security","ref":"references/automated-weapons.md"}
{"kind":"fitness","id":"landscape_scoring","formula":"survivors / tests_run per dimension","example":"error-handling: 5/8 = 0.625 (high fitness - keep probing)"}
{"kind":"allocation","fitness":">0.7","challengers":"6+","status":"HEMORRHAGING"}
{"kind":"allocation","fitness":">0.5","challengers":"5","status":"HIGH PRESSURE"}
{"kind":"allocation","fitness":">0.3","challengers":"4","status":"CONTESTED"}
{"kind":"allocation","fitness":">0.1","challengers":"3","status":"PROBING"}
{"kind":"allocation","fitness":"0","challengers":"2","status":"COOLING (double-tap, never single)"}
{"kind":"allocation","fitness":"exhausted 3+ gens","challengers":"0","status":"DORMANT (reawaken every 5 gens)"}
{"kind":"allocation","multiplier":"1.0x + 0.5x per 2 generations","status":"Escalating pressure"}
{"kind":"equilibrium","criteria":"3 consecutive zero-survivor generations","note":"The Queen never truly rests - dormant dims reawaken every 5 gens"}
{"kind":"severity","id":"CRITICAL","criteria":"Core workflow command returns wrong exit code","landscape":"dimension.fitness = 1.0 (maximum pressure)","examples":"0 on error, non-0 on success"}
{"kind":"severity","id":"MAJOR","criteria":"Documented command fails or produces wrong output","landscape":"dimension.fitness += 0.3","examples":"Verified by grep/diff, not AI"}
{"kind":"severity","id":"MINOR","criteria":"Output doesn't match documented format","landscape":"dimension.fitness += 0.1","examples":"Verified by pattern match, not AI"}
{"kind":"severity","id":"OBSERVATION","criteria":"AI-only judgment (no deterministic check)","landscape":"Not added to done_when","note":"Only severity that relies on AI"}
{"kind":"rule","id":"defeat_all_predecessors","text":"New code must beat entire lineage. Use lineage-replay to verify.","enforcement":"Every done_when from every generation must pass"}
{"kind":"rule","id":"anti_stagnation","text":"Dormant dimensions reawaken every 5 generations. The Queen never truly rests."}
{"kind":"rule","id":"severity_escalation","text":"3+ consecutive survivors in a dimension auto-promote severity. Persistent wounds become critical."}
{"kind":"rule","id":"double_tap","text":"Never send just 1 challenger to a cooling dimension. Always 2+. Confirm the kill."}
{"kind":"rule","id":"carnage_tracking","text":"Monitor kill rate and lethality per dimension. The codebase must constantly defend itself."}
{"kind":"anti_pattern","id":"ai_decides_test","problem":"Nondeterministic, unreproducible","fix":"Exit code comparison only"}
{"kind":"anti_pattern","id":"ai_scores_landscape","problem":"Subjective, varies between runs","fix":"survivors / tests_run ratio"}
{"kind":"anti_pattern","id":"ai_decides_stop","problem":"No convergence guarantee","fix":"Equilibrium: 0 survivors for 3 consecutive gens"}
{"kind":"anti_pattern","id":"ai_gates_transitions","problem":"Bypassable, inconsistent","fix":"assert-can-* functions in liza"}
{"kind":"anti_pattern","id":"tests_not_persisted","problem":"Lost between sessions","fix":"done_when in blackboard YAML"}
{"kind":"anti_pattern","id":"ai_judges_regression","problem":"Flaky, depends on prompt","fix":"validate runs all checks, exit code comparison"}
{"kind":"verdict","id":"crown_defended","criteria":"ALL done_when pass","output":"CROWN DEFENDED"}
{"kind":"verdict","id":"crown_contested","criteria":"MAJOR/MINOR survivors exist, all done_when pass","output":"CROWN CONTESTED"}
{"kind":"verdict","id":"crown_forfeit","criteria":"CRITICAL survivors exist","output":"CROWN FORFEIT"}
{"kind":"phase","id":"probe","step":0,"ai":["Read README, --help, source code. Discover promises."],"code":["Register each promise as done_when via task-add-check"]}
{"kind":"phase","id":"evolve","step":"N","ai":["Generate 3-10 test commands based on landscape"],"code":["gen-start, execute commands, gen-survivor or gen-discard"]}
{"kind":"phase","id":"regress","step":"N+1","ai":[],"code":["coder-submit, validate, lineage-replay, carnage"]}
{"kind":"phase","id":"automate","step":"parallel","ai":[],"code":["spec-mine, quality-gate, fowler-review, mutate"]}
{"kind":"ref","file":"references/protocol.md","use":"Full execution protocol with examples"}
{"kind":"ref","file":"references/automated-weapons.md","use":"mutate, spec-mine, quality-gate, fowler-review details"}

L="$HOME/.claude/skills/red-queen/liza-advanced.nu"
nu $L init
nu $L task-add drq-session --spec_ref README.md
nu $L task-add-check drq-session "myapp --help" --expect_exit=0
nu $L claim drq-session red-queen

nu $L gen-start drq-session
# AI generates test commands...
# Execute each, exit code decides survivor vs discard
if [ $? -eq 0 ]; then  # BUG: should have failed
  nu $L gen-survivor drq-session "dim" "cmd" --severity CRITICAL
  br create --title "[Red Queen] CRITICAL: <finding>" --type=bug
else
  nu $L gen-discard drq-session "dim"
fi
nu $L landscape drq-session
nu $L coder-submit drq-session red-queen
nu $L validate drq-session