{
  "published": "2026-10-07",
  "scope": "One pull request, one review per setting (two at medium). A screen, not a leaderboard: do not generalize the counts.",
  "pull_request": {
    "url": "https://github.com/pgup-ai/jbot-review/pull/286",
    "files": 9,
    "known_issues": 2,
    "known_issues_source": "The two shell-parsing bugs fixed afterwards in https://github.com/pgup-ai/jbot-review/pull/287: and-chain-break and wrapped-cd"
  },
  "setup": {
    "reviewer": "J-Bot Review local mode, OpenCode engine, context-pack preset, verification off, 10 concurrent sessions",
    "haiku_route": "anthropic/claude-haiku-5-5 (direct Anthropic API)",
    "comparison_route": "deepseek/deepseek-flash (direct DeepSeek API, DeepSeek V4.1 Flash) at low effort",
    "cost_source": "Per-session cost reported by OpenCode at list prices, to four decimals. The session costs and the run total are each rounded on their own, so a split can differ from its total by $0.0001.",
    "guideline_pass_note": "In these runs the guideline-compliance pass ran at the same effort as the main review. Current J-Bot runs it at low."
  },
  "definitions": {
    "confirmed_bug": "A defect we reproduced against the code, or one a later merged PR fixed.",
    "rule_finding": "A valid violation of the repository's written rules in AGENTS.md.",
    "medium": "No effort sent. Anthropic documents medium as Haiku 5.5's default, identical to omitting the parameter.",
    "found_by": "Run ids from the runs list."
  },
  "runs": [
    {
      "id": "low",
      "model": "Claude Haiku 5.5",
      "effort": "low",
      "wall_seconds": 29,
      "cost_usd": 0.015,
      "main_cost_usd": 0.008,
      "guideline_cost_usd": 0.0071,
      "tool_calls_main": 0,
      "tool_calls_guideline": 1,
      "reasoning_tokens_main": 5160,
      "confirmed_bugs": 1,
      "rule_findings": 0,
      "known_issues_found": 0
    },
    {
      "id": "medium-1",
      "model": "Claude Haiku 5.5",
      "effort": "medium",
      "wall_seconds": 79,
      "cost_usd": 0.022,
      "main_cost_usd": 0.0164,
      "guideline_cost_usd": 0.0056,
      "tool_calls_main": 3,
      "tool_calls_guideline": 0,
      "reasoning_tokens_main": 15296,
      "confirmed_bugs": 0,
      "rule_findings": 0,
      "known_issues_found": 0
    },
    {
      "id": "medium-2",
      "model": "Claude Haiku 5.5",
      "effort": "medium",
      "wall_seconds": 92,
      "cost_usd": 0.028,
      "main_cost_usd": 0.0207,
      "guideline_cost_usd": 0.0073,
      "tool_calls_main": 8,
      "tool_calls_guideline": 1,
      "reasoning_tokens_main": 17326,
      "confirmed_bugs": 1,
      "rule_findings": 1,
      "known_issues_found": 0
    },
    {
      "id": "high",
      "model": "Claude Haiku 5.5",
      "effort": "high",
      "wall_seconds": 163,
      "cost_usd": 0.0455,
      "main_cost_usd": 0.029,
      "guideline_cost_usd": 0.0165,
      "tool_calls_main": 7,
      "tool_calls_guideline": 6,
      "reasoning_tokens_main": 32628,
      "confirmed_bugs": 2,
      "rule_findings": 1,
      "known_issues_found": 0
    },
    {
      "id": "xhigh",
      "model": "Claude Haiku 5.5",
      "effort": "xhigh",
      "wall_seconds": 438,
      "cost_usd": 0.3304,
      "main_cost_usd": 0.2808,
      "guideline_cost_usd": 0.0496,
      "tool_calls_main": 32,
      "tool_calls_guideline": 8,
      "reasoning_tokens_main": 95222,
      "confirmed_bugs": 2,
      "rule_findings": 1,
      "known_issues_found": 0
    },
    {
      "id": "max",
      "model": "Claude Haiku 5.5",
      "effort": "max",
      "wall_seconds": 751,
      "cost_usd": 1.2476,
      "main_cost_usd": 0.6667,
      "guideline_cost_usd": 0.5809,
      "tool_calls_main": 45,
      "tool_calls_guideline": 37,
      "reasoning_tokens_main": 163579,
      "confirmed_bugs": 3,
      "rule_findings": 1,
      "known_issues_found": 1
    },
    {
      "id": "deepseek-low",
      "model": "DeepSeek V4.1 Flash",
      "effort": "low",
      "wall_seconds": 265,
      "cost_usd": 0.0817,
      "main_cost_usd": 0.0487,
      "guideline_cost_usd": 0.033,
      "tool_calls_main": 49,
      "tool_calls_guideline": 20,
      "reasoning_tokens_main": 49967,
      "confirmed_bugs": 1,
      "rule_findings": 0,
      "known_issues_found": 0
    }
  ],
  "bugs": [
    {
      "id": "and-chain-break",
      "summary": "An unknown && segment stopped parsing the whole shell command, dropping reads after a later ;",
      "status": "fixed in #287",
      "found_by": [
        "max"
      ],
      "known_issue": true
    },
    {
      "id": "wrapped-cd",
      "summary": "A wrapped cd (command cd, builtin cd, pushd, popd) or a cd after the first segment left later read paths unknown",
      "status": "fixed in #287",
      "known_issue": true,
      "found_by": []
    },
    {
      "id": "telemetry-workspace",
      "summary": "Three backends computed tool path identities without the workspace, so one file counted twice",
      "status": "fixed in #293",
      "found_by": [
        "low",
        "medium-2",
        "high",
        "xhigh",
        "max",
        "deepseek-low"
      ],
      "known_issue": false
    },
    {
      "id": "quoted-or-operator",
      "summary": "A quoted || in a grep pattern ended the search-pattern scan",
      "status": "fixed in #293",
      "found_by": [
        "max"
      ],
      "known_issue": false
    },
    {
      "id": "quoted-dollar",
      "summary": "A single-quoted span inside double quotes hid a $ from the expansion guard",
      "status": "open",
      "found_by": [
        "high",
        "xhigh"
      ],
      "known_issue": false
    }
  ]
}