Hook insertion in add_repo() used hand-rolled precedent/successor logic against a static _HOOK_ORDER list. This replaces it with pipeweld's Adder, which models the same ordering constraints as a series-parallel graph — laying groundwork for repo-aware insertion and prek priority support.

Changes

• func.py — Adder: Added force_linear setting that post-processes the solution to flatten any Parallel nodes into a deterministic linear Series. Added _extract_ordered_steps() and _linearize_component() helpers internal to the pipeweld layer.
• hooks.py — add_repo(): Build a Series pipeline from existing hook IDs, derive prerequisites/postrequisites sets from _HOOK_ORDER, and delegate insertion positioning to Adder.add() with force_linear=True. The predecessor is read directly from the flat solution.
• test_func.py: Unit tests for _extract_ordered_steps, _linearize_component, and Adder with force_linear=True (16 new tests)
• test_hooks.py: Integration tests exercising pipeweld-based insertion (prerequisite present, postrequisite only, mixed non-dependent/postrequisite)

Before/After

# Before: manual list slicing
precedents = _HOOK_ORDER[:hook_idx]
successors = _HOOK_ORDER[hook_idx + 1:]
existing_precedents = [h for h in existing_hooks if h in precedents]
# ... 15 lines of if/elif/else to find last_precedent

# After: pipeweld solver
adder = Adder(
    pipeline=series(*existing_hooks),
    step=hook_config.id,
    prerequisites=set(_HOOK_ORDER[:hook_idx]),
    postrequisites=set(_HOOK_ORDER[hook_idx + 1:]),
    force_linear=True,
)
result = adder.add()
# Solution is a flat Series of strings — read predecessor directly
flat = result.solution.root
idx = flat.index(hook_config.id)
predecessor = flat[idx - 1] if idx > 0 else None

All existing tests pass unchanged — behavioral compatibility is preserved for current hook orderings.

⌨️ Start Copilot coding agent tasks without leaving your editor — available in VS Code, Visual Studio, JetBrains IDEs and Eclipse.