Serialize the full OptKnob spec into the IR + fix KNOB_MUTATE

[akashgit]

What

Two changes that make KNOB_MUTATE (the highest-weight outer-loop operator, 0.25) actually contribute to the swarm:

1. knob_specs in the IR — Workflow now carries a full knob_specs table (kind / node_id / default / bounds / expandable / expansion_hint / description) beside knob_values / knob_bounds. to_dict() serializes it, from_dict() reconstructs it, and Package.compile() populates it.

2. KNOB_MUTATE survives mutation + novelty filtering — two defects silently discarded every knob-only mutation:

   • _deep_copy_workflow dropped knob_specs (and task / declared_capabilities), so children lost their OptKnob declarations.
   • structural_hash excluded knob_values and NoveltyFilter.is_novel rejected any workflow whose graph-edit-distance to an archived candidate was below threshold — even when only knob values differed. Now structural_hash includes knob_values, and the near-duplicate check only applies when knob_values match.

Why

Without these, KNOB_MUTATE was selected ~25% of the time but its results were always rejected as "duplicates", so the operator was dead weight.

Tests

• pytest tests/test_outer_loop/test_similarity.py tests/test_outer_loop/test_mutations.py tests/test_outer_loop/test_engine.py tests/test_outer_loop/test_seed_diversity.py tests/test_outer_loop/test_reflector.py — 174 passed.
• A downstream harness confirms all 8 operators now fire and the archive/reflection/evaluate path works end to end.

[github-actions]

Sentrux Quality Report

Absolute

Scanning ....
[scan] git ls-files: 651 total, 637 kept, 14 dropped (ext:14, meta:0, big:0)
[build_project_map] 637 files, 108 unique dirs, 100 cache misses, 2.6ms
[resolve] 1296 resolved, 1680 unresolved (of 2976 total specs)
[resolve_imports] project_map 2.7ms, suffix_idx 0.8ms, suffix_resolve 12.4ms, total 15.9ms
[build_graphs] 637 files | maps 1.6ms, imports 16.1ms, calls+inherit 4.9ms, total 22.6ms | 1295 import, 10472 call, 4 inherit edges
sentrux check — 3 rules checked

Quality: 4415

✗ [Error] max_cc: 9 function(s) exceed max cyclomatic complexity of 30
    factory/workflow/executor.py:_execute_data (cc=53)
    factory/outer_loop/reflector.py:_extract_eval_patterns (cc=46)
    factory/cli/_ceo_helpers.py:_validate_ceo_flags (cc=43)
    factory/cli/_ceo_helpers.py:_execute_ceo (cc=43)
    factory/outer_loop/mutations.py:_try_mutation (cc=40)
    examples/chess_evolve.py:main (cc=33)
    factory/cli/_task_builder.py:_build_ceo_task (cc=32)
    factory/outer_loop/reflector.py:_llm_reflect (cc=32)
    factory/outer_loop/mutations.py:validate_and_repair (cc=31)

✗ 1 violation(s) found

Diff (vs base branch)

Scanning ....
[scan] git ls-files: 651 total, 637 kept, 14 dropped (ext:14, meta:0, big:0)
[build_project_map] 637 files, 108 unique dirs, 100 cache misses, 2.7ms
[resolve] 1296 resolved, 1680 unresolved (of 2976 total specs)
[resolve_imports] project_map 2.8ms, suffix_idx 0.9ms, suffix_resolve 14.7ms, total 18.4ms
[build_graphs] 637 files | maps 1.7ms, imports 18.6ms, calls+inherit 5.5ms, total 25.8ms | 1295 import, 10472 call, 4 inherit edges
sentrux gate — structural regression check

Quality:      4415 -> 4415
Coupling:     0.81 → 0.81
Cycles:       3 → 3
God files:    3 → 3

Distance from Main Sequence: 0.35

✓ No degradation detected

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 83.74%. Comparing base (4f83d6c) to head (4295cf2).
⚠️ Report is 4 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1499      +/-   ##
==========================================
+ Coverage   83.62%   83.74%   +0.12%     
==========================================
  Files         225      225              
  Lines       25447    25757     +310     
  Branches     4138     4209      +71     
==========================================
+ Hits        21280    21571     +291     
- Misses       3202     3204       +2     
- Partials      965      982      +17     

☔ 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.