test(source-control): add a plugin eval suite for resolve-conflicts

.changes/source-control/test-resolve-conflicts-plugin-eval-suite-ee7a239e.md

@@ -0,0 +1,7 @@
+---
+bump: patch
+---
+
+### Added
+
+- **A `claude plugin eval` suite for `/source-control:resolve-conflicts`.** Seven cases tagged `row22`, each scaffolding a git repository that is stopped in a real conflict: a two-stop rebase, a six-file merge where the prompt asks for `--ours` or abort, a merge with a conflict that has no markers, a cherry-pick, a merge whose incoming tip never touched the file, a revert, and a merge with only "sort it out" as the prompt. Most graders are deterministic: the resolved file contents, history read before the first edit, tests run before the operation concludes, and the concluding reflog line. They also cover the "It's working if" checks from the archived upstream skill page. Three cases add one model-judged grader each, and every grader carries pass and fail samples. The skill body is unchanged.

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/case.yaml

@@ -0,0 +1,4 @@
+schema_version: "1.1"
+name: resolve-conflicts-bare-prompt-recovers-intent-from-history
+context:
+  scaffold_script: scaffold.sh

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/history-before-edit.md

@@ -0,0 +1,5 @@
+---
+type: regex
+pattern: '^(?:(?!"name":"(?:Edit|Write|MultiEdit|NotebookEdit)"|"command":")[\s\S]|"command":"(?:(?!(?:\bsed\s+-i|\bperl\s+-\w*i|(?<![\d&])>(?!&)|\b(?:cp|mv|tee)\s)(?:[^"\\;&|]|\\(?!n).)*?(?<!\w)client\.py\b|\b(?:checkout|restore)\s+(?:\S+\s+)*?--(?:ours|theirs)\b)(?:[^"\\]|\\.))*")*?"command":"(?:(?:(?!(?:\bsed\s+-i|\bperl\s+-\w*i|(?<![\d&])>(?!&)|\b(?:cp|mv|tee)\s)(?:[^"\\;&|]|\\(?!n).)*?(?<!\w)client\.py\b|\b(?:checkout|restore)\s+(?:\S+\s+)*?--(?:ours|theirs)\b)(?:[^"\\]|\\.))*?(?:[;&|]|\\n))?\s*git\s+(?:-C\s+\S+\s+)?(?:--no-pager\s+)?(?:log|show)\b'
+target: trace
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/intent-composed.md

@@ -0,0 +1,5 @@
+---
+type: regex
+pattern: '^(?![\s\S]*legacy_token)(?=[\s\S]*region)'
+target: { source: file, path: api/client.py }
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/no-abort.md

@@ -0,0 +1,8 @@
+---
+type: tool_used
+tool: Bash
+input_match: '--abort\b'
+min: 0
+max: 0
+arm: both
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/no-markers-client.md

@@ -0,0 +1,7 @@
+---
+type: regex
+pattern: '^(?:<{7}|={7}|>{7})'
+flags: m
+match: not_contains
+target: { source: file, path: api/client.py }
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/skill-fired.md

@@ -0,0 +1,5 @@
+---
+type: tool_used
+tool: Skill
+input_match: '"skill"\s*:\s*"(?:source-control:)?resolve-conflicts"'
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/up-checks-before-conclude.md

@@ -0,0 +1,5 @@
+---
+type: tool_order
+before: { tool: Bash, input_match: 'unittest|pytest|python3?\s+\S*test_\w+\.py' }
+after: { tool: Bash, input_match: 'merge\s+--continue|git\s+commit\b' }
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/up-intents-and-drops-named.md

@@ -0,0 +1,9 @@
+---
+type: llm
+---
+
+Main's commit removed `legacy_token` from `fetch_orders` because the /orders endpoint rejects token query auth with a 400 (SEC-12). The feature commit added a `region` filter for the EU dashboard (ORD-5).
+
+PASS if the reply states each side's intent as the commit history gives it (the removal and its reason; the region filter and its purpose) and says that `legacy_token` is gone from the result because main removed it on purpose.
+
+FAIL if the reply describes the resolution only in terms of the diff or the conflict markers, keeps `legacy_token` alongside `region`, or drops `region`.

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/up-operation-complete.md

@@ -0,0 +1,5 @@
+---
+type: regex
+pattern: 'commit \(merge\)'
+target: { source: file, path: .git/logs/HEAD }
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/graders/up-quotes-history.md

@@ -0,0 +1,6 @@
+---
+type: regex
+pattern: '^\{"type":"assistant".*"type":"text","text":"(?:[^"\\]|\\.)*?(?:SEC-12|\b400\b|stopped accepting|token query auth)'
+flags: mi
+target: trace
+---

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/prompt.md

@@ -0,0 +1,11 @@
+---
+description: "Bare 'sort it out' merge prompt: only the commit message says legacy_token was removed on purpose; keep region, drop legacy_token, say why, run tests, conclude"
+tags: [resolve-conflicts, source-control, row22, upstream-working-if]
+runs: 3
+max_turns: 40
+timeout_seconds: 900
+allowed_tools: [Read, Glob, Grep, Skill, Bash, Edit]
+expected_outcome: "api/client.py keeps region and has no legacy_token, no markers, history was read before the edit, the reply cites SEC-12 or the 400, tests ran before the merge was concluded, and the merge commit exists"
+---
+
+git merge main stopped with conflicts. Sort it out and finish the merge.

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/samples/history-before-edit.json

@@ -0,0 +1,75 @@
+{
+  "pass": [
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"BASE=$(git merge-base HEAD MERGE_HEAD); git log -p \\\"$BASE\\\"..MERGE_HEAD -- client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Write\",\"input\":{\"file_path\":\"/w/client.py\",\"content\":\"y\\n\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_2\",\"name\":\"Bash\",\"input\":{\"command\":\"git add client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_2\",\"content\":\"ok\"}]}}"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git status\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git show abc123\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_2\",\"name\":\"Edit\",\"input\":{\"file_path\":\"/w/client.py\",\"old_string\":\"x\",\"new_string\":\"y\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_2\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_3\",\"name\":\"Bash\",\"input\":{\"command\":\"git -C /w add client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_3\",\"content\":\"ok\"}]}}"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git --no-pager log --oneline -5 -- client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}",
+      "why": "history read; nothing edited yet"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"python3 -m unittest > test-output.log 2>&1\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "redirect to an unrelated file is not an edit of the conflicted file"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git status 2>&1 >/dev/null\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline 2>&1\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "stderr and null redirects are not file writes"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"cp notes.txt backup.txt && git log --oneline -- client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}",
+      "why": "unrelated write chained with a history read of the conflicted file"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git log -- client.py && cat theirs > client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}",
+      "why": "history read first; the rewrite comes after it in the same call"
+    }
+  ],
+  "fail": [
+    {
+      "answer": ""
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Write\",\"input\":{\"file_path\":\"/w/client.py\",\"content\":\"y\\n\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_2\",\"name\":\"Bash\",\"input\":{\"command\":\"git add client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_2\",\"content\":\"ok\"}]}}",
+      "why": "rewrote the file before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Edit\",\"input\":{\"file_path\":\"/w/client.py\",\"old_string\":\"x\",\"new_string\":\"y\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git show abc123\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "edited before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"sed -i s/x/y/ client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log -p -- client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "rewrote the file with Bash before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git checkout --theirs client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "took a side before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git diff\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Write\",\"input\":{\"file_path\":\"/w/client.py\",\"content\":\"y\\n\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_2\",\"name\":\"Bash\",\"input\":{\"command\":\"git add client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_2\",\"content\":\"ok\"}]}}",
+      "why": "diff only, no history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"git status\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_9\",\"content\":\"try git log next\"}]}}",
+      "why": "git log only in tool output"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"cat theirs.py > client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "overwrote the file with a redirect before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"cp theirs.py client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_1\",\"name\":\"Bash\",\"input\":{\"command\":\"git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_1\",\"content\":\"ok\"}]}}",
+      "why": "copied over the file before reading history"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"cat theirs > client.py && git log -- client.py\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}",
+      "why": "rewrote the file earlier in the same call as the history read"
+    },
+    {
+      "answer": "{\"type\":\"system\",\"subtype\":\"init\",\"model\":\"m\"}\n{\"type\":\"assistant\",\"message\":{\"role\":\"assistant\",\"content\":[{\"type\":\"tool_use\",\"id\":\"toolu_0\",\"name\":\"Bash\",\"input\":{\"command\":\"sed -i s/x/y/ client.py; git log --oneline\",\"description\":\"run\"}}]}}\n{\"type\":\"user\",\"message\":{\"content\":[{\"type\":\"tool_result\",\"tool_use_id\":\"toolu_0\",\"content\":\"ok\"}]}}",
+      "why": "edited with sed earlier in the same call as the history read"
+    }
+  ]
+}

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/samples/intent-composed.json

@@ -0,0 +1,15 @@
+{
+  "pass": [
+    {
+      "answer": "def fetch_orders(session, customer_id, region=None):\n    params = {\"customer\": customer_id}\n    if region:\n        params[\"region\"] = region\n    return session.get(\"/orders\", params=params)\n"
+    }
+  ],
+  "fail": [
+    {
+      "answer": "def fetch_orders(session, customer_id, legacy_token=None, region=None):\n    params = {\"customer\": customer_id}\n    if region:\n        params[\"region\"] = region\n    if legacy_token:\n        params[\"token\"] = legacy_token\n"
+    },
+    {
+      "answer": "def fetch_orders(session, customer_id):\n    params = {\"customer\": customer_id}\n"
+    }
+  ]
+}

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/samples/no-abort.json

@@ -0,0 +1,39 @@
+{
+  "pass": [
+    {
+      "answer": []
+    },
+    {
+      "answer": [
+        {
+          "tool": "Bash",
+          "input": {
+            "command": "git status | head -4"
+          }
+        }
+      ]
+    }
+  ],
+  "fail": [
+    {
+      "answer": [
+        {
+          "tool": "Bash",
+          "input": {
+            "command": "git merge --abort"
+          }
+        }
+      ]
+    },
+    {
+      "answer": [
+        {
+          "tool": "Bash",
+          "input": {
+            "command": "git rebase --abort && git status"
+          }
+        }
+      ]
+    }
+  ]
+}

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/samples/no-markers-client.json

@@ -0,0 +1,19 @@
+{
+  "pass": [
+    {
+      "answer": "x = 1\ny = 2\n"
+    },
+    {
+      "answer": "# a ======= rule in a comment is fine mid-line ======= \n"
+    }
+  ],
+  "fail": [
+    {
+      "answer": "<<<<<<< HEAD\nx = 1\n=======\nx = 2\n>>>>>>> main\n"
+    },
+    {
+      "answer": "x = 1\n=======\n",
+      "why": "a leftover separator"
+    }
+  ]
+}

plugins/source-control/evals/resolve-conflicts-bare-prompt-recovers-intent-from-history/samples/skill-fired.json

@@ -0,0 +1,46 @@
+{
+  "pass": [
+    {
+      "answer": [
+        {
+          "tool": "Skill",
+          "input": {
+            "skill": "source-control:resolve-conflicts"
+          }
+        }
+      ]
+    },
+    {
+      "answer": [
+        {
+          "tool": "Bash",
+          "input": {
+            "command": "git status"
+          }
+        },
+        {
+          "tool": "Skill",
+          "input": {
+            "skill": "resolve-conflicts"
+          }
+        }
+      ]
+    }
+  ],
+  "fail": [
+    {
+      "answer": []
+    },
+    {
+      "answer": [
+        {
+          "tool": "Skill",
+          "input": {
+            "skill": "source-control:commit"
+          }
+        }
+      ],
+      "why": "wrong skill"
+    }
+  ]
+}