{ "tokens_out": 3995, "outcome": "pass", "suggested_fix": null, "cycle_checks": null, "fault": { "owner": null, "status": "resolved", "reason": null, "evidence": [] }, "runtime_revision": null, "agent_execution": "native_agent", "time_to_working_s": null, "usage_complete": true, "runtime": null, "architecture": null, "local_or_hosted": null, "agent_interfaces_note": null, "platform_details": null, "platforms_claimed": null, "maintenance": null, "evidence_strength": null, "purpose": "initial", "topic": null, "needs_which": null, "tested_with": null, "native_bundle": null, "agent_runtime": "claude-code", "maintenance_note": null, "runtime_version": null, "does_what_it_says": null, "commit": "25dfbb1ddc6de0f2c4d5bc38b6267e351461b52b", "failure_origin": null, "stuck_step": null, "repo_url": null, "needs": null, "checks": [ { "check": "operation", "status": "PASS", "details": { "working": true, "argv": [ "demo" ], "kind": "cli", "returncode": 0, "stdout": "The demo bot is a tiny network that learned to copy a simple scripted player. Let's look inside it.\n\nStep 1 of 4 - recording 12 games (60 s each) against a ball chaser in RocketSim\n game 1/12: blue 0 - orange 1 (300 moments so far, 0 s)\n game 2/12: blue 0 - orange 2 (600 moments so far, 1 s)\n game 3/12: blue 1 - orange 0 (900 moments so far, 1 s)\n game 4/12: blue 1 - orange 1 (1200 moments so far, 1 s)\n game 5/12: blue 1 - orange 1 (1500 moments so far, 2 s)\n game 6/12: blue 0 - orange 0 (1800 moments so far, 2 s)\n game 7/12: blue 1 - orange 1 (2100 moments so far, 3 s)\n game 8/12: blue 0 - orange 0 (2400 moments so far, 3 s)\n game 9/12: blue 1 - orange 1 (2700 moments so far, 3 s)\n[excerpt truncated]", "stderr": "", "needs_credentials": false, "needs_browser": false, "scope": "product operation" } } ], "open_issues": null, "agent_runtime_version": "2.1.300 (Claude Code)", "verifies_run_id": null, "total_s": 185.86910605430603, "license": null, "tool_calls": 9, "model_version": "claude-haiku-5-5", "example_checks": [], "mcp": null, "other_model": null, "first_use_s": null, "native_tool_calls": 9, "last_commit_at": null, "source": "https://github.com/adventure-gpt/cds-interpret", "models_agree": null, "maintenance_verdict": null, "cost_usd": 0.00629457, "safety": null, "answered_issues": null, "what_failed": null, "steps": [ "cd && ls -la && cat README.md 2>/dev/null | head -200", "echo CDP=$CDP_URL $; cat | head -60; which node npm python3 uv pip3 2>&1; python3 --version", "cd && (python3 -m venv .venv 2>&1 || (sudo apt-get update -qq && sudo apt-get install -y -qq python3-venv >/dev/null 2>&1 && python3 -m venv .venv)) && .venv/bin/pip install -q --upgrade pip 2>&1 | tail -3; .venv/bin/pip install -q torch --index-url https://download.pytorch.org/whl/cpu 2>&1 | tail -5; echo EXIT=$?", "cd && rm -rf .venv && (sudo -n apt-get update -qq 2>&1 | tail -2; sudo -n DEBIAN_FRONTEND=noninteractive apt-get install -y -qq python3.12-venv 2>&1 | tail -3); python3 -m venv .venv && .venv/bin/python -m pip --version", "cd && .venv/bin/pip install -q --upgrade pip 2>&1 | tail -3; .venv/bin/pip install -q torch --index-url https://download.pytorch.org/whl/cpu 2>&1 | tail -5; echo TORCH_EXIT=${PIPESTATUS[0]}", "cd && .venv/bin/pip install -q \".[rlgym,rlbot]\" 2>&1 | tail -15; echo EXIT=${PIPESTATUS[0]}; ls .venv/bin | grep -i cds", "cd && export NODE_USE_ENV_PROXY=1 && timeout 1500 .venv/bin/cds-interpret demo > /tmp/demo.log 2>&1; echo EXIT=$?; tail -40 /tmp/demo.log; ls -la cds-demo 2>/dev/null | head", "cd && mkdir -p .coldpass && cat > .coldpass/probe.json <<'EOF'\n{\"kind\":\"cli\",\"argv\":[\"\",\"demo\"],\"cwd\":\"\",\"env\":{\"NODE_USE_ENV_PROXY\":\"1\"}}\nEOF\ncat > .coldpass/report.json <<'EOF'\n{\"outcome\":\"working\",\n\"operation\":\"cds-interpret demo: records a built-in bot in RocketSim, probes each layer, trains a sparse autoencoder, runs causal edits, and writes cds-demo/report.html\",\n\"operation_scope\":\"Ran the full demo on CPU torch with rlgym and rlbot extras. Not exercised: live RLBot overlay (needs a running Rocket League/RLBot match), user's own bot, browser viewing of report.html (no browser used).\",\n\"steps\":[\"python3 -m venv .venv (after installing python3.12-venv via apt)\",\n\"pip install --upgrade pip\",\n\"pip install torch --index-url https://download.pytorch.org/whl/cpu\",\n\"pip install \\\".[rlgym,rlbot]\\\" (from the pinned checkout)\",\n\".venv/bin/cds-interpret demo (exit 0, wrote cds-demo/report.html)\"],\n\"deviations\":[\"venv creation failed with ensurepip unavailable; ran apt-get install python3.12-venv, then recreated .venv\",\n\"README says pip install \\\"cds-interpret[rlgym,rlbot]\\\" from PyPI; installed from the pinned checkout instead with pip install \\\".[rlgym,rlbot]\\\" so the code matches commit 25dfbb1\",\n\"Did not run unit tests or lint, per task instructions\"],\n\"stuck_step\":null,\"main_blocker\":null,\"suggested_fix\":null}\nEOF\n= python3 -I --validate", "cd && rm -rf cds-demo && = python3 -I --validate && ls -la .coldpass && pgrep -fa cds-interpret || echo \"no leftover processes\"" ], "platform_conflicts": null, "claims_note": null, "maintained_note": null, "repo": "adventure-gpt/cds-interpret", "service_tier": null, "maintained": null, "source_pin": null, "operation_scope": "product operation", "reasoning": "low", "claims": [], "platform": "linux", "cost_to_try": null, "backend": null, "tool_transport": "mcp-assigned-vm", "hardware": null, "agent_interface": null, "profile": { "model": "claude-haiku-5-5", "reasoning": "low" }, "date": "2026-10-18T18:03:49.498838+00:00", "kind": "repo", "resources": { "cpus": 2, "memory_mib": 4096, "disk_mib": 32768 }, "error_excerpt": null, "pass": true, "evidence_version": null, "source_revision": null, "run_id": "adventure-gpt--cds-interpret-claude-haiku-5-5-1792346443629717215", "last_release_at": null, "furthest_model": null, "model": "claude-haiku-5-5", "what_it_did": "cds-interpret demo: records a built-in bot in RocketSim, probes each layer, trains a sparse autoencoder, runs causal edits, and writes cds-demo/report.html", "sandbox_source": null, "other_run_id": null, "hosting": null, "tested_commit": null, "guest_setup": null, "mcp_context": null, "agent_interfaces": [], "setup_identity": null, "works": null, "install_s": null, "fix_id": null, "tokens_in": 128186, "tested_at": null, "test_version": "readme-2e7d3cb1b49a8213de0d5c33ce91d6c77432809ec683ffa1501492f502faa935", "recovery": { "source": "legacy-oct16-18", "original_run_id": "adventure-gpt--cds-interpret-claude-haiku-5-5-1792346443629717215", "historical": true, "ledger_sha256": "c5e6a833c182c1e2aecf69d0bd0020aec63c65d98f3e933e4b32232340312736" }, "transcript_url": "/transcripts/adventure-gpt--cds-interpret-claude-haiku-5-5-1792346443629717215.txt", "report_url": "/reports/adventure-gpt--cds-interpret-claude-haiku-5-5-1792346443629717215.json" }