fix: use detected default branch instead of hardcoded main in CEO workflow

[osilkin98]

Closes #730

Changes

• Replaced 7 hardcoded main references in factory/agents/prompts/ceo.md with $TARGET_BRANCH (PR targets, baseline SHA lookups, guard violation checkout)
• Added TARGET_BRANCH resolution instruction near the start of the CEO prompt so the variable is set from .factory/config.json each cycle
• Changed _build_ceo_task in factory/cli.py to always inject the resolved target branch (via _read_target_branch), not only when --branch is explicitly passed
• Added error handling in _read_target_branch to gracefully fall back to "main" if detect_default_branch fails

[github-actions]

Sentrux Quality Report

Running sentrux check...
Scanning ....
[scan] git ls-files: 259 total, 249 kept, 10 dropped (ext:10, meta:0, big:0)
[build_project_map] 249 files, 42 unique dirs, 38 cache misses, 4.3ms
[resolve] 376 resolved, 659 unresolved (of 1035 total specs)
[resolve_imports] project_map 4.3ms, suffix_idx 0.5ms, suffix_resolve 10.5ms, total 15.3ms
[build_graphs] 249 files | maps 1.0ms, imports 15.4ms, calls+inherit 3.0ms, total 19.5ms | 375 import, 3843 call, 0 inherit edges
sentrux check — 2 rules checked

Quality: 4967

✗ [Error] max_cc: 4 function(s) exceed max cyclomatic complexity of 30
    factory/cli.py:cmd_ceo (cc=78)
    factory/cli.py:_welcome_wizard (cc=39)
    factory/cli.py:cmd_run (cc=36)
    factory/study.py:study_project_local (cc=36)

✗ 1 violation(s) found

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 87.84%. Comparing base (600cad3) to head (fa19b46).

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #731   +/-   ##
=======================================
  Coverage   87.83%   87.84%           
=======================================
  Files          70       70           
  Lines       10699    10704    +5     
=======================================
+ Hits         9398     9403    +5     
  Misses       1301     1301           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[osilkin98]

✅ Factory Review: KEEP

Verdict: KEEP
Reason: All hardcoded main references replaced with $TARGET_BRANCH in ceo.md; CLI unconditionally injects resolved branch; guard check clean; 176 smoke tests pass; no critical issues

Experiment: #3
Hypothesis: Replace hardcoded main in CEO prompt and always inject resolved target branch in CLI

Score Comparison

Metric	Value
Before	0.0000
After	0.0000
Delta	+0.0000
Threshold	0.6000

Guard Checks

Check	Result
eval_immutable	✅ PASS
scope	✅ PASS
git_clean	✅ PASS

Precheck Gate

Guard check passed clean. No eval scores available (eval_command empty). Smoke tests: 176 passed.

Code Review Notes

• try/except around detect_default_branch correctly handles missing git/non-repo dirs with main fallback
• TARGET_BRANCH resolution in ceo.md provides shell-level variable for all git commands
• 9 hardcoded main refs in git commands replaced consistently
• Unconditional Target Branch injection in _build_ceo_task ensures CEO always knows the branch
• No remaining hardcoded main in git command patterns (grep verified)

---

Posted by Factory CEO

[osilkin98]

✅ Factory Review: KEEP

Verdict: KEEP
Reason: All 9 hardcoded main refs replaced with $TARGET_BRANCH, CLI always injects resolved branch

Experiment: #3
Hypothesis: Replace hardcoded main in CEO prompt and always inject resolved target branch in CLI

Score Comparison

Metric	Value
Before	0.6166
After	0.6213
Delta	+0.0047
Threshold	0.6000

Guard Checks

Check	Result
scope	✅ PASS
eval_immutable	✅ PASS

---

Posted by Factory CEO

[osilkin98]

Closing this as it was posted on version 1 factory