Add "minimal" pipeline template for trivial single-file fixes

.claude/daemon-config.json

@@ -16,7 +16,7 @@
   "pipeline_template": "autonomous",
   "auto_template": false,
   "last_optimization": {
-    "timestamp": "2026-03-08T15:49:14Z",
+    "timestamp": "2026-03-09T12:06:32Z",
     "adjustments": "compound_quality enabled (CFR 25% > 20%); merge stage recommended (deploy freq 3.0/week); "
   }
 }

.claude/intelligence-cache.json

@@ -32,7 +32,7 @@
       "result": {
         "raw_response": "Mock claude response"
       },
-      "timestamp": 1772984580,
+      "timestamp": 1773056069,
       "ttl": 3600
     },
     "efd5524ad1244fd20344bf3ff5d95233": {
@@ -263,7 +263,7 @@
           }
         }
       ],
-      "timestamp": 1772984911,
+      "timestamp": 1773059374,
       "ttl": 1800
     },
     "fdbb7bc340414cf938cecd4b790fa9e2": {
@@ -647,6 +647,31 @@
       },
       "timestamp": 1772936833,
       "ttl": 3600
+    },
+    "293f94c5d80cb1596386597316f534d3": {
+      "result": {
+        "complexity": 5,
+        "risk_level": "medium",
+        "success_probability": 78,
+        "recommended_template": "fast",
+        "issue_type": "testing",
+        "key_risks": [
+          "False positives on legitimate repeated work or transient failures",
+          "State hashing overhead in tight iteration loop",
+          "Integration with existing loop restart and error injection logic",
+          "Edge cases in file change detection and git status parsing"
+        ],
+        "implementation_hints": [
+          "Hash (git status + changed files + last 3 lines of test output) per iteration using sha256sum",
+          "Track hashes in memory array with iteration metadata (number, timestamp, hash)",
+          "Implement early return before test execution when stall detected to save cost",
+          "Add --no-stall-detection flag and environment variable override",
+          "Test with mock iterations that produce identical hashes, different hashes, and mixed patterns",
+          "Write detailed stall reason to error-log.jsonl including hash sequence and iteration range"
+        ]
+      },
+      "timestamp": 1773058311,
+      "ttl": 3600
     }
   }
 }

.claude/loop-logs/iteration-1.log

@@ -1,11 +1 @@
-Perfect! The goal has been fully achieved. The shipwright CLI now has a `hello` command that outputs "hello world". The implementation:
-
-1. ✅ Added the `hello` command to the help text
-2. ✅ Implemented the command case statement in the main router
-3. ✅ Tested the command works correctly
-4. ✅ All tests pass
-5. ✅ Committed with a descriptive message
-
-The command is now available as `shipwright hello` (and with aliases `sw hello` and `cct hello`).
-
-LOOP_COMPLETE
+{"type":"result","subtype":"success","is_error":false,"duration_ms":529352,"duration_api_ms":415316,"num_turns":65,"result":"All tests pass. Here's what was implemented:\n\n## Summary\n\n**New `minimal` pipeline template** (`templates/pipelines/minimal.json`):\n- Stages: intake → build → test → PR (compound_quality, plan, review, deploy all disabled)\n- Model: `haiku` (cheapest/fastest)\n- Max build iterations: 3 (vs 10 for fast/hotfix)\n- All intelligence disabled (no adversarial, architecture, or simulation)\n- No quality gates\n\n**Integration points updated:**\n- **Project detection** (`scripts/lib/project-detect.sh`): Recommends `minimal` for projects with ≤5 source files, no CI, no deploy\n- **Daemon triage** (`scripts/lib/daemon-triage.sh`): Score-based selection uses `minimal` for score ≥90 (very trivial issues)\n- **Upgrade** (`scripts/sw-upgrade.sh`): Added to file distribution list\n- **CLAUDE.md** (both project and global): Updated pipeline template tables\n- **Tests**: Updated project-detect and daemon-triage tests with new assertions\n\nLOOP_COMPLETE","stop_reason":"end_turn","session_id":"2f82583e-c851-41f2-9d8a-bb99e0145eab","total_cost_usd":2.91068575,"usage":{"input_tokens":56,"cache_creation_input_tokens":93351,"cache_read_input_tokens":3929906,"output_tokens":13527,"server_tool_use":{"web_search_requests":0,"web_fetch_requests":0},"service_tier":"standard","cache_creation":{"ephemeral_1h_input_tokens":93351,"ephemeral_5m_input_tokens":0},"inference_geo":"","iterations":[],"speed":"standard"},"modelUsage":{"claude-opus-4-6":{"inputTokens":56,"outputTokens":13527,"cacheReadInputTokens":3929906,"cacheCreationInputTokens":93351,"webSearchRequests":0,"costUSD":2.8868517499999995,"contextWindow":200000,"maxOutputTokens":32000},"claude-haiku-4-5-20251001":{"inputTokens":20379,"outputTokens":691,"cacheReadInputTokens":0,"cacheCreationInputTokens":0,"webSearchRequests":0,"costUSD":0.023834000000000004,"contextWindow":200000,"maxOutputTokens":32000}},"permission_denials":[],"fast_mode_state":"off","uuid":"ff624623-f744-46a5-a04c-78711b713e43"}

.claude/loop-logs/iteration-2.log

@@ -1 +1 @@
-Invalid API key · Fix external API key
+{"type":"result","subtype":"success","is_error":false,"duration_ms":531817,"duration_api_ms":338403,"num_turns":52,"result":"All 58 pipeline tests and 19 smoke tests pass. The root cause was that parent pipeline environment variables (`PIPELINE_TEMPLATE=fast`, `ARTIFACTS_DIR`, `SCRIPT_DIR`, etc.) leaked into the test subprocess, causing it to load the wrong template and skip the plan/review stages.\n\nLOOP_COMPLETE","stop_reason":"end_turn","session_id":"264e0e4e-4a83-471b-ba74-817cddd4e9fd","total_cost_usd":2.932504249999999,"usage":{"input_tokens":54,"cache_creation_input_tokens":91271,"cache_read_input_tokens":4004087,"output_tokens":13535,"server_tool_use":{"web_search_requests":0,"web_fetch_requests":0},"service_tier":"standard","cache_creation":{"ephemeral_1h_input_tokens":91271,"ephemeral_5m_input_tokens":0},"inference_geo":"","iterations":[],"speed":"standard"},"modelUsage":{"claude-opus-4-6":{"inputTokens":54,"outputTokens":13535,"cacheReadInputTokens":4004087,"cacheCreationInputTokens":91271,"webSearchRequests":0,"costUSD":2.91113225,"contextWindow":200000,"maxOutputTokens":32000},"claude-haiku-4-5-20251001":{"inputTokens":17952,"outputTokens":684,"cacheReadInputTokens":0,"cacheCreationInputTokens":0,"webSearchRequests":0,"costUSD":0.021372,"contextWindow":200000,"maxOutputTokens":32000}},"permission_denials":[],"fast_mode_state":"off","uuid":"7f9062f5-4a88-4801-8a6a-91a29d660ef7"}