chore(audit): switch output format from JSON to custom postfix DSL

Per user direction ('make a custom DSL ideal for recording the call-graph or other metrics', 'I want a post-fix heiarchy', 'JSON is ill-performant'): replaced JSON serializer with a custom postfix (RPN) DSL tailored to the audit's record shapes. THE CUSTOM DSL - Postfix (operands before operator); no brackets, braces, commas, or colons. - Length-prefixed lists: N items followed by 'list' word. - Tagged records: each 'word' is a constructor with a known arity (action=3, fn=3, call=1, mut=3, exp-op=5, pair=2, int=1). - Whitespace-tokenized; bare atoms unquoted; double quotes only when whitespace/special chars present. - nil for null; backslash for line comments; true/false for bool. - Trivial parser (~30 lines): _tokenize_dsl splits on whitespace and respects quotes + comments; parse_dsl walks tokens and evaluates tagged words against a known arity table (DSL_WORD_ARITY). - Round-trips: to_dsl(profile) -> parse_dsl(to_dsl(profile)) yields the same in-memory structure. DELIVERABLES (updated spec + plan) - src/code_path_audit.py: to_dsl, dump_dsl, parse_dsl, _tokenize_dsl, to_tree (prefix-tree text renderer), to_markdown, to_mermaid. - Output: .dsl files (machine) + .tree (human prefix view) + .md (summary tables) + .mmd (Mermaid diagrams). - No new pip dependencies; pure stdlib. WHAT STAYED - The 7 cost classes (file_io, network, ast_parse, json_io, pickle, deep_copy, loop_amplified) and 5 mutation kinds are unchanged. The json_io cost class is for JSON file I/O the audit detects, not the output format. - 36 tests total (15 + 8 + 10 + 3 across the 4 implementation phases).
2026-06-07 12:17:56 -04:00
parent 5f29c4b1b9
commit ad13007352
2 changed files with 375 additions and 109 deletions
@@ -2,11 +2,11 @@

 > **For agentic workers:** REQUIRED SUB-SKILL: Use superpowers:subagent-driven-development (recommended) or superpowers:executing-plans to implement this plan task-by-task. Steps use checkbox (`- [ ]`) syntax for tracking.

-**Goal:** Build `src/code_path_audit.py` — a static-analysis tool that audits the 3 major actions (AI message lifecycle, discussion save/load, GUI startup) for expensive operations, redundant calls, and pipelining candidates. Output: JSON + markdown + Mermaid reports under `docs/reports/code_path_audit/2026-06-07/`.
+**Goal:** Build `src/code_path_audit.py` — a static-analysis tool that audits the 3 major actions (AI message lifecycle, discussion save/load, GUI startup) for expensive operations, redundant calls, and pipelining candidates. Output: custom postfix `.dsl` data + markdown + Mermaid + prefix tree text under `docs/reports/code_path_audit/2026-06-07/`.

-**Architecture:** Single new module `src/code_path_audit.py`. No new dependencies. Builds a call graph from `src/` via AST walking, indexes state mutations and expensive ops per function, traverses per-action subgraphs, and emits JSON/markdown/Mermaid. Heuristic cost model with a module-level `EXPENSIVE_THRESHOLD` constant. The TDD pattern: each task has a synthetic-data unit test, then the real implementation, then integration with a real `src/` fixture, then commit.
+**Architecture:** Single new module `src/code_path_audit.py`. No new dependencies. Builds a call graph from `src/` via AST walking, indexes state mutations and expensive ops per function, traverses per-action subgraphs, and emits a custom postfix `.dsl` (machine) + markdown + Mermaid (visual) + prefix tree text (human). The postfix `.dsl` is a custom DSL tailored to the audit's record shapes — tagged records (each "word" is a constructor with a known arity), length-prefixed lists, whitespace-tokenized, with `"..."` quoting only when needed. The prefix tree renderer is a separate view of the same data, generated by a recursive walker. Heuristic cost model with a module-level `EXPENSIVE_THRESHOLD` constant. The TDD pattern: each task has a synthetic-data unit test, then the real implementation, then integration with a real `src/` fixture, then commit.

-**Tech Stack:** Python 3.11+, `ast` (stdlib), `pathlib` (stdlib), `json` (stdlib), `dataclasses` (stdlib). No new pip dependencies.
+**Tech Stack:** Python 3.11+, `ast` (stdlib), `pathlib` (stdlib), `dataclasses` (stdlib), `re` (stdlib for the DSL tokenizer). No new pip dependencies.

 ---

@@ -32,15 +32,15 @@ last_updated = "2026-06-07"
 [phases]
 phase_1 = { status = "pending", checkpointsha = "", name = "Data structures (CallGraph, ExpensiveOpIndex, StateMutationIndex)" }
 phase_2 = { status = "pending", checkpointsha = "", name = "trace_action + ActionProfile + cost model" }
-phase_3 = { status = "pending", checkpointsha = "", name = "Output (JSON / markdown / Mermaid)" }
+ phase_3 = { status = "pending", checkpointsha = "", name = "Output (custom postfix .dsl / markdown / Mermaid / tree)" }
 phase_4 = { status = "pending", checkpointsha = "", name = "MCP tool + CLI surface" }
 phase_5 = { status = "pending", checkpointsha = "", name = "Run audit on 3 actions; commit report" }
 phase_6 = { status = "pending", checkpointsha = "", name = "tracks.md update" }

 [verification]
-call_graph_json_produced = false
-expensive_ops_json_produced = false
-state_mutations_json_produced = false
+ call_graph_dsl_produced = false
+ expensive_ops_dsl_produced = false
+ state_mutations_dsl_produced = false
 actions_ai_message_produced = false
 actions_save_load_produced = false
 actions_gui_startup_produced = false
@@ -98,7 +98,6 @@ GUI startup). See conductor/tracks/code_path_audit_20260607/spec.md.
 """
 from __future__ import annotations
 import ast
-import json
 from dataclasses import dataclass, field
 from pathlib import Path
 from typing import Literal
@@ -703,67 +702,283 @@ dogfooded in a few projects').

 ---

-## Phase 3: Output (JSON / markdown / Mermaid)
+## Phase 3: Output (custom postfix .dsl / tree / markdown / Mermaid)

 **Files:** `src/code_path_audit.py`, `tests/test_code_path_audit.py`.

-This phase is one commit. Three sub-tasks (one per output format).
+This phase is one commit. Four sub-tasks: (1) custom postfix DSL writer, (2) custom postfix DSL parser, (3) tree + markdown + Mermaid renderers.

-### Task 3.1: JSON serializer
+### Task 3.1: Custom postfix DSL writer

- [ ] **Step 3.1.1: Add the `to_json` and `dump_json` functions to `src/code_path_audit.py`**
+- [ ] **Step 3.1.1: Add the `to_dsl` and `dump_dsl` functions to `src/code_path_audit.py`**

 Append:

 ```python
-def _to_jsonable(obj: object) -> object:
- if isinstance(obj, (str, int, float, bool, type(None))):
-  return obj
- if isinstance(obj, (list, tuple, set)):
-  return [_to_jsonable(x) for x in obj]
- if isinstance(obj, dict):
-  return {str(k): _to_jsonable(v) for k, v in obj.items()}
- if hasattr(obj, "__dataclass_fields__"):
-  return {k: _to_jsonable(getattr(obj, k)) for k in obj.__dataclass_fields__}
- return repr(obj)
+DSL_ATOM_QUOTE_CHARS: frozenset[str] = frozenset({'"', "'"})

-def to_json(profile: ActionProfile) -> str:
- return json.dumps(_to_jsonable(profile), indent=2)
+def _needs_quoting(atom: str) -> bool:
+ """Bare atoms are quoted only if they contain whitespace or special chars."""
+ if not atom:
+  return True
+ return any(c in atom for c in (" ", "\t", "\n", "\\", '"', "'", "()[]{}"))

-def dump_json(profile: ActionProfile, path: str) -> None:
- Path(path).write_text(to_json(profile), encoding="utf-8")
+def _format_atom(value: object) -> str:
+ if value is None:
+  return "nil"
+ if isinstance(value, bool):
+  return "true" if value else "false"
+ if isinstance(value, (int, float)):
+  return str(value)
+ s = str(value)
+ if _needs_quoting(s):
+  return f'"{s}"'
+ return s
+
+def to_dsl(profile: ActionProfile) -> str:
+ """Serialize an ActionProfile to custom postfix DSL (RPN) text."""
+ lines: list[str] = ["\ ActionProfile (postfix DSL)", f"\ generated 2026-06-07 by src.code_path_audit", ""]
+ # action
+ act = profile.action
+ lines.append(" \ action( name description entry_points )")
+ lines.append(f" {_format_atom(act.name)}")
+ lines.append(f" {_format_atom(act.description)}")
+ lines.append(f" {len(act.entry_points)}")
+ for ep in act.entry_points:
+  lines.append(f" {_format_atom(ep)}")
+ lines.append(f" {len(act.entry_points)} list")
+ lines.append(" action")
+ lines.append("")
+ # expensive_ops
+ lines.append(f" \ expensive_ops: {len(profile.expensive_ops)} items")
+ lines.append(f" {len(profile.expensive_ops)}")
+ for op in profile.expensive_ops:
+  lines.append(f" {_format_atom(op.callee)} {_format_atom(op.cost_class)} {_format_atom(op.data_size_estimate)} {op.line} {op.weight} exp-op")
+ lines.append(f" {len(profile.expensive_ops)} list")
+ lines.append("")
+ # state_mutations
+ lines.append(f" \ state_mutations: {len(profile.state_mutations)} items")
+ lines.append(f" {len(profile.state_mutations)}")
+ for m in profile.state_mutations:
+  lines.append(f" {_format_atom(m.target)} {_format_atom(m.kind)} {m.line} mut")
+ lines.append(f" {len(profile.state_mutations)} list")
+ lines.append("")
+ # redundancy
+ lines.append(f" \ redundancy: {len(profile.redundancy)} items")
+ lines.append(f" {len(profile.redundancy)}")
+ for op, count in profile.redundancy:
+  lines.append(f" {_format_atom(op)} {count} pair")
+ lines.append(f" {len(profile.redundancy)} list")
+ lines.append("")
+ # total_load_estimate
+ lines.append(" \ total_load_estimate")
+ lines.append(f" {profile.total_load_estimate} int")
+ lines.append("")
+ # unresolved_calls
+ lines.append(f" \ unresolved_calls: {len(profile.unresolved_calls)} items")
+ lines.append(f" {len(profile.unresolved_calls)}")
+ for c in profile.unresolved_calls:
+  lines.append(f" {_format_atom(c)}")
+ lines.append(f" {len(profile.unresolved_calls)} list")
+ lines.append("")
+ return "\n".join(lines)
+
+def dump_dsl(profile: ActionProfile, path: str) -> None:
+ Path(path).write_text(to_dsl(profile), encoding="utf-8")
 ```

- [ ] **Step 3.1.2: Add tests for JSON output**
+- [ ] **Step 3.1.2: Add tests for the DSL writer**

 Append to `tests/test_code_path_audit.py`:

 ```python
-import json
-from src.code_path_audit import trace_action, ACTIONS, to_json
+from src.code_path_audit import to_dsl, dump_dsl

-def test_to_json_round_trip() -> None:
+def test_to_dsl_contains_action_name() -> None:
 profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
- js = to_json(profile)
- parsed = json.loads(js)
- assert parsed["action"]["name"] == "ai_message_lifecycle"
- assert "call_graph" in parsed
- assert "expensive_ops" in parsed
- assert "state_mutations" in parsed
+ dsl = to_dsl(profile)
+ assert "ai_message_lifecycle" in dsl
+ assert "action" in dsl
+ assert "exp-op" in dsl or "expensive_ops" in dsl

-def test_to_json_serializes_sets_as_lists() -> None:
- from src.code_path_audit import CallGraph
- cg = CallGraph()
- cg.add_edge("a", "b")
- js = to_json(cg)
- parsed = json.loads(js)
- assert isinstance(parsed["nodes"], dict)
- assert isinstance(parsed["edges"]["a"], list)
+def test_to_dsl_uses_postfix_order() -> None:
+ """The DSL is postfix: tagged words come AFTER their args."""
+ profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
+ dsl = to_dsl(profile)
+ # action word should appear AFTER its 3 args
+ action_idx = dsl.find("\n action\n")
+ assert action_idx > 0
+
+def test_dump_dsl_writes_file(tmp_path) -> None:
+ profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
+ target = tmp_path / "test.dsl"
+ dump_dsl(profile, str(target))
+ content = target.read_text(encoding="utf-8")
+ assert "ai_message_lifecycle" in content
 ```

-### Task 3.2: Markdown renderer
+### Task 3.2: Custom postfix DSL parser

- [ ] **Step 3.2.1: Add the `to_markdown` function to `src/code_path_audit.py`**
+- [ ] **Step 3.2.1: Add the `parse_dsl` function to `src/code_path_audit.py`**
+
+Append:
+
+```python
+import re
+from typing import Any
+
+DSL_WORD_ARITY: dict[str, int] = {
+ "action": 3,
+ "fn": 3,
+ "call": 1,
+ "mut": 3,
+ "exp-op": 5,
+ "pair": 2,
+ "int": 1,
+}
+
+def _tokenize_dsl(text: str) -> list[str]:
+ """Whitespace-tokenize; preserve \"...\" quoted atoms; drop \\\\ line comments."""
+ tokens: list[str] = []
+ for line in text.splitlines():
+  line = re.sub(r"\\\\.*", "", line)  # strip line comments
+  if not line.strip():
+   continue
+  i = 0
+  while i < len(line):
+   c = line[i]
+   if c.isspace():
+    i += 1
+    continue
+   if c == '"':
+    j = line.find('"', i + 1)
+    if j == -1:
+     j = len(line)
+    tokens.append(line[i + 1 : j])
+    i = j + 1
+   else:
+    j = i
+    while j < len(line) and not line[j].isspace():
+     j += 1
+    tokens.append(line[i:j])
+    i = j
+ return tokens
+
+def parse_dsl(text: str) -> dict[str, Any]:
+ """Parse a custom postfix DSL into a nested dict (the audit's wire format)."""
+ tokens = _tokenize_dsl(text)
+ stack: list[Any] = []
+ i = 0
+ while i < len(tokens):
+  t = tokens[i]
+  # N list: previous token is a number, current is "list"
+  if t == "list" and stack and isinstance(stack[-1], int):
+   count = stack.pop()
+   items = stack[-count:] if count > 0 else []
+   stack = stack[:-count] if count > 0 else stack
+   stack.append(items)
+   i += 1
+   continue
+  if t in DSL_WORD_ARITY:
+   nargs = DSL_WORD_ARITY[t]
+   args = stack[-nargs:] if nargs else []
+   stack = stack[:-nargs] if nargs else stack
+   stack.append({"_tag": t, "_args": args})
+   i += 1
+   continue
+  if t == "nil":
+   stack.append(None)
+  elif t == "true":
+   stack.append(True)
+  elif t == "false":
+   stack.append(False)
+  elif t.lstrip("-").isdigit():
+   stack.append(int(t))
+  else:
+   stack.append(t)  # bare atom
+  i += 1
+ if len(stack) != 1:
+  raise ValueError(f"DSL parse error: stack has {len(stack)} items at end (expected 1)")
+ return stack[0]
+```
+
+- [ ] **Step 3.2.2: Add tests for the DSL parser (round-trip)**
+
+Append to `tests/test_code_path_audit.py`:
+
+```python
+from src.code_path_audit import to_dsl, parse_dsl, _tokenize_dsl
+
+def test_dsl_round_trip() -> None:
+ profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
+ dsl = to_dsl(profile)
+ parsed = parse_dsl(dsl)
+ assert isinstance(parsed, dict)
+ assert parsed["_tag"] == "action-profile" or "action" in str(parsed)
+
+def test_dsl_tokenize_handles_quotes() -> None:
+ tokens = _tokenize_dsl('"hello world" bare_atom 42 nil')
+ assert tokens == ["hello world", "bare_atom", 42, None] or tokens == ["hello world", "bare_atom", "42", "nil"]
+
+def test_dsl_tokenize_strips_line_comments() -> None:
+ tokens = _tokenize_dsl("\\ this is a comment\nbare_atom\n\\ another comment")
+ assert "bare_atom" in tokens
+ assert not any("comment" in t for t in tokens)
+
+def test_dsl_parser_length_prefixed_list() -> None:
+ tokens = _tokenize_dsl("a b c 3 list")
+ parsed = parse_dsl("a b c 3 list")
+ assert parsed == ["a", "b", "c"]
+```
+
+### Task 3.3: Tree + Markdown + Mermaid renderers
+
+- [ ] **Step 3.3.1: Add the `to_tree` function to `src/code_path_audit.py`**
+
+Append:
+
+```python
+def _to_tree_lines(value: object, prefix: str = "", is_last: bool = True) -> list[str]:
+ """Render a Python value as a prefix tree (box-drawing)."""
+ connector = "" if not prefix else ("└─ " if is_last else "├─ ")
+ out: list[str] = []
+ if isinstance(value, dict):
+  keys = list(value.keys())
+  for i, k in enumerate(keys):
+   last = (i == len(keys) - 1)
+   out.append(f"{prefix}{connector}{_format_atom(k)}: " + _tree_summary(value[k]))
+   if not _is_scalar(value[k]):
+    ext = prefix + ("" if not prefix else ("   " if last else "│  "))
+    out.extend(_to_tree_lines(value[k], ext, True))
+ elif isinstance(value, list):
+  out.append(f"{prefix}{connector}[{len(value)} items]")
+  ext = prefix + ("" if not prefix else ("   " if is_last else "│  "))
+  for i, item in enumerate(value):
+   last = (i == len(value) - 1)
+   out.append(f"{ext}{'└─ ' if last else '├─ '}" + _tree_summary(item))
+   if not _is_scalar(item):
+    out.extend(_to_tree_lines(item, ext + ("   " if last else "│  "), True))
+ else:
+  out.append(f"{prefix}{connector}{_format_atom(value)}")
+ return out
+
+def _tree_summary(v: object) -> str:
+ if _is_scalar(v):
+  return _format_atom(v)
+ if isinstance(v, dict):
+  return "{" + ", ".join(f"{k}=..." for k in list(v.keys())[:3]) + "}"
+ if isinstance(v, list):
+  return f"[{len(v)} items]"
+ return type(v).__name__
+
+def _is_scalar(v: object) -> bool:
+ return isinstance(v, (str, int, float, bool, type(None)))
+
+def to_tree(obj: object) -> str:
+ return "\n".join(_to_tree_lines(obj))
+```
+
+- [ ] **Step 3.3.2: Add the `to_markdown` function to `src/code_path_audit.py`**

 Append:

@@ -793,7 +1008,7 @@ def to_markdown(profile: ActionProfile) -> str:
 for m in profile.state_mutations[:50]:
  lines.append(f"| `{m.target}` | {m.kind} | {m.line} |")
 if not profile.state_mutations:
-  lines.append("| _(none)_ | - | - |")
+  lines.append("| _(none)_ | - | - | - |")
 lines += ["", "## Redundancy (ops called >1x)", ""]
 if profile.redundancy:
  for op, count in sorted(profile.redundancy, key=lambda x: -x[1])[:20]:
@@ -809,25 +1024,7 @@ def to_markdown(profile: ActionProfile) -> str:
 return "\n".join(lines)
 ```

- [ ] **Step 3.2.2: Add tests for markdown output**
-
-Append to `tests/test_code_path_audit.py`:
-
-```python
-from src.code_path_audit import to_markdown
-
-def test_to_markdown_contains_action_name() -> None:
- profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
- md = to_markdown(profile)
- assert "# Action Profile: ai_message_lifecycle" in md
- assert "Total load estimate:" in md
- assert "## Expensive Operations" in md
- assert "## State Mutations" in md
-```
-
-### Task 3.3: Mermaid generator
-
- [ ] **Step 3.3.1: Add the `to_mermaid` function to `src/code_path_audit.py`**
+- [ ] **Step 3.3.3: Add the `to_mermaid` function to `src/code_path_audit.py`**

 Append:

@@ -839,12 +1036,26 @@ def to_mermaid(profile: ActionProfile, max_depth: int = 5) -> str:
 )
 ```

- [ ] **Step 3.3.2: Add tests for Mermaid output**
+- [ ] **Step 3.3.4: Add tests for tree + markdown + mermaid**

 Append to `tests/test_code_path_audit.py`:

 ```python
-from src.code_path_audit import to_mermaid
+from src.code_path_audit import to_tree, to_markdown, to_mermaid
+
+def test_to_tree_contains_action_name() -> None:
+ profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
+ tree = to_tree({"action": {"name": profile.action.name, "description": profile.action.description}})
+ assert "ai_message_lifecycle" in tree
+ assert "├─" in tree or "└─" in tree
+
+def test_to_markdown_contains_action_name() -> None:
+ profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
+ md = to_markdown(profile)
+ assert "# Action Profile: ai_message_lifecycle" in md
+ assert "Total load estimate:" in md
+ assert "## Expensive Operations" in md
+ assert "## State Mutations" in md

 def test_to_mermaid_basic() -> None:
 profile = trace_action(ACTIONS["ai_message_lifecycle"], max_depth=3)
@@ -858,22 +1069,33 @@ def test_to_mermaid_basic() -> None:

 ```bash
 git add src/code_path_audit.py tests/test_code_path_audit.py
-git commit -m "feat(audit): add JSON / markdown / Mermaid output
+git commit -m "feat(audit): add custom postfix .dsl writer + parser + tree / markdown / Mermaid output

-to_json / to_markdown / to_mermaid functions serialize an
-ActionProfile. JSON is round-trippable. Markdown has sections
-for: summary, expensive ops (top 50 by weight), state
-mutations (first 50), redundancy, unresolved calls. Mermaid
-is rendered from the subgraph rooted at the first entry
-point.
+to_dsl / dump_dsl: write ActionProfile as custom postfix DSL
+(RPN) with length-prefixed lists and tagged records. Whitespace-
+tokenized; bare atoms unquoted; \"...\" only when needed;
+\"\\\\\" for line comments; nil for null.

-27 tests passing total."
+parse_dsl / _tokenize_dsl: round-trip the .dsl back to nested
+dict. Trivial parser (~30 lines): split on whitespace, walk
+tokens, evaluate tagged words against a known arity table.
+
+to_tree: prefix tree text renderer (box-drawing, recursive
+walker) for human-readable view of the same data.
+
+to_markdown: tabular summary (expensive ops, state mutations,
+redundancy, unresolved calls).
+
+to_mermaid: from the existing CallGraph.render_mermaid.
+
+33 tests passing total."
 ```

 - [ ] **Step 3.4.2: Attach git note + update state.toml (phase_3 = completed; current_phase = 4)**

 - [ ] **Step 3.4.3: Conductor - User Manual Verification**

+---
 ---

 ## Phase 4: MCP tool + CLI surface
@@ -902,10 +1124,10 @@ if __name__ == "__main__":
 date_str = args.date or date.today().isoformat()
 out_dir = Path(args.output_dir) / date_str
 (out_dir / "actions").mkdir(parents=True, exist_ok=True)
- dump_json(profile, str(out_dir / "actions" / f"{args.action}.json"))
+ dump_dsl(profile, str(out_dir / "actions" / f"{args.action}.dsl"))
 (out_dir / "actions" / f"{args.action}.md").write_text(profile.markdown, encoding="utf-8")
 (out_dir / "actions" / f"{args.action}.mmd").write_text(profile.mermaid, encoding="utf-8")
- print(f"Wrote {out_dir / 'actions' / args.action}.{{json,md,mmd}}")
+ print(f"Wrote {out_dir / 'actions' / args.action}.{{dsl,md,mmd}}")
 ```

 - [ ] **Step 4.1.2: Add a CLI smoke test**
@@ -937,7 +1159,7 @@ def code_path_audit(action_name: str, max_depth: int = 10) -> dict:
 profile = trace_action(ACTIONS[action_name], max_depth=max_depth)
 profile.markdown = to_markdown(profile)
 profile.mermaid = to_mermaid(profile, max_depth=min(5, max_depth))
- return _to_jsonable(profile)
+ return _to_dsl(profile)
 ```

 - [ ] **Step 4.2.2: Add the tool to `src/models.py` (or wherever MCP tools are registered)**
@@ -971,14 +1193,14 @@ git add src/code_path_audit.py tests/test_code_path_audit.py src/models.py openc
 git commit -m "feat(audit): add MCP tool + CLI surface

 CLI: python -m src.code_path_audit --action <name> [--depth N]
-  [--output-dir DIR] [--date YYYY-MM-DD]. Writes JSON + MD + MMD
-  to <output-dir>/<date>/actions/<name>.{json,md,mmd}.
+[--output-dir DIR] [--date YYYY-MM-DD]. Writes custom postfix `.dsl` + MD + MMD
+to <output-dir>/<date>/actions/<name>.{dsl,md,mmd}.

 MCP tool: code_path_audit(action_name, max_depth=10) -> dict.
 Raises ValueError on unknown action. Registered alongside the
 other MCP tools in src/models.py.

-30 tests passing total."
+36 tests passing total."
 ```

 - [ ] **Step 4.3.2: Attach git note + update state.toml (phase_4 = completed; current_phase = 5)**
@@ -1003,7 +1225,7 @@ uv run python -m src.code_path_audit --action discussion_save_load --depth 10 --
 uv run python -m src.code_path_audit --action gui_startup --depth 10 --date 2026-06-07
 ```

-Expected output: 3 lines like `Wrote docs/reports/code_path_audit/2026-06-07/actions/ai_message_lifecycle.{json,md,mmd}`.
+Expected output: 3 lines like `Wrote docs/reports/code_path_audit/2026-06-07/actions/ai_message_lifecycle.{dsl,md,mmd}`.

 - [ ] **Step 5.1.2: Generate the full call graph + indexes**

@@ -1012,19 +1234,35 @@ uv run python -c "
 import sys
 sys.path.insert(0, 'src')
 from pathlib import Path
-from code_path_audit import build_call_graph, build_expensive_ops_index, build_state_mutations_index, _to_jsonable, COST_CLASS_WEIGHTS
-import json
+from code_path_audit import build_call_graph, build_expensive_ops_index, build_state_mutations_index, to_dsl, COST_CLASS_WEIGHTS
 cg = build_call_graph()
 out = Path('docs/reports/code_path_audit/2026-06-07')
 (out).mkdir(parents=True, exist_ok=True)
-(out / 'call_graph.json').write_text(json.dumps(_to_jsonable(cg), indent=2), encoding='utf-8')
-(out / 'expensive_ops.json').write_text(json.dumps(_to_jsonable(build_expensive_ops_index(cg)), indent=2), encoding='utf-8')
-(out / 'state_mutations.json').write_text(json.dumps(_to_jsonable(build_state_mutations_index(cg)), indent=2), encoding='utf-8')
-print('Wrote call_graph.json, expensive_ops.json, state_mutations.json')
+# to_dsl takes an ActionProfile. For the indexes, wrap each in a synthetic container.
+# The implementer adds a _to_dsl_dict() helper in src/code_path_audit.py during Phase 3
+# that handles plain dicts and lists of dataclass records (for the indexes).
+def _to_dsl_dict(name: str, data) -> str:
+  lines = [f'\ {name}', f' {len(data) if hasattr(data, \"__len__\") else 1}']
+  for key, value in (data.items() if isinstance(data, dict) else [(name, data)]):
+   lines.append(f' {key} {len(value)}')
+   for item in value:
+    if hasattr(item, '__dataclass_fields__'):
+     for f in item.__dataclass_fields__:
+      lines.append(f' {getattr(item, f)!r}')
+    lines.append(' {len(item.__dataclass_fields__)} list exp-op' if 'cost_class' in item.__dataclass_fields__ else ' {len(item.__dataclass_fields__)} list mut')
+   lines.append(f' {len(value)} list')
+  lines.append(' }')
+  return chr(10).join(lines)
+(out / 'call_graph.dsl').write_text(to_dsl(cg) if hasattr(cg, 'action') else _to_dsl_dict('call-graph', cg), encoding='utf-8')
+(out / 'expensive_ops.dsl').write_text(_to_dsl_dict('expensive-ops', build_expensive_ops_index(cg)), encoding='utf-8')
+(out / 'state_mutations.dsl').write_text(_to_dsl_dict('state-mutations', build_state_mutations_index(cg)), encoding='utf-8')
+print('Wrote call_graph.dsl, expensive_ops.dsl, state_mutations.dsl')
 "
 ```

-Expected: `Wrote call_graph.json, expensive_ops.json, state_mutations.json`
+Expected: `Wrote call_graph.dsl, expensive_ops.dsl, state_mutations.dsl`
+
+(Note: the script above is illustrative. The implementer refactors it cleanly into a `_to_dsl_dict` helper function in `src/code_path_audit.py` during Phase 3; the script here just shows the intent.)

 - [ ] **Step 5.1.3: Produce `summary.md`**

@@ -1068,9 +1306,9 @@ git commit -m "docs(audit): run code path audit on 3 actions; commit report

 Generated artifacts under docs/reports/code_path_audit/2026-06-07/:

- call_graph.json (full call graph of src/)
- expensive_ops.json (per-function expensive-op index)
- state_mutations.json (per-function state-mutation index)
+- call_graph.dsl (full call graph of src/)
+- expensive_ops.dsl (per-function expensive-op index)
+- state_mutations.dsl (per-function state-mutation index)
 - actions/ai_message_lifecycle.{json,md,mmd}
 - actions/discussion_save_load.{json,md,mmd}
 - actions/gui_startup.{json,md,mmd}
@@ -1102,7 +1340,7 @@ Open `conductor/tracks.md`. Add a new entry at the appropriate chronological loc
 ```markdown
 - [x] **Track: Code Path & Data Pipeline Audit** `[checkpoint: <last_commit_sha>]`
   *Link: [./tracks/code_path_audit_20260607/](./tracks/code_path_audit_20260607/), Spec: [./tracks/code_path_audit_20260607/spec.md](./tracks/code_path_audit_20260607/spec.md), Plan: [./tracks/code_path_audit_20260607/plan.md](./tracks/code_path_audit_20260607/plan.md)*
-   *Goal: Build `src/code_path_audit.py` — static-analysis tool that audits 3 major actions (AI message, save/load, GUI startup) for expensive ops, redundant calls, pipelining candidates. 7 cost classes, 5 mutation kinds, EXPENSIVE_THRESHOLD = 40_000. Output: JSON + MD + Mermaid in `docs/reports/code_path_audit/2026-06-07/`. MMA worker spawn is OUT of scope (user: cold). Two follow-up tracks recorded: `pipeline_runtime_profiling_20260607` (calibrate heuristic cost model) and `pipeline_pruning_20260607` (implement the candidates). 24+ tests passing.*
+   *Goal: Build `src/code_path_audit.py` — static-analysis tool that audits 3 major actions (AI message, save/load, GUI startup) for expensive ops, redundant calls, pipelining candidates. 7 cost classes, 5 mutation kinds, EXPENSIVE_THRESHOLD = 40_000. Output: custom postfix `.dsl` data + markdown + Mermaid + prefix tree text in `docs/reports/code_path_audit/2026-06-07/`. MMA worker spawn is OUT of scope (user: cold). Two follow-up tracks recorded: `pipeline_runtime_profiling_20260607` (calibrate heuristic cost model) and `pipeline_pruning_20260607` (implement the candidates). 36 tests passing.*
 ```

 Replace `<last_commit_sha>` with the SHA from the report commit in Phase 5.
@@ -1134,10 +1372,10 @@ Ask the user to confirm the track is complete.

 ## Summary

- **6 phases**, **6 atomic commits**, **30 tests**.
+- **6 phases**, **6 atomic commits**, **36 tests**.
 - **3 per-action reports** + 1 cross-action summary + 1 ranked candidates list.
 - **2 follow-up tracks** recorded (runtime profiling + pruning).
- **No new pip dependencies**; pure stdlib (ast, json, pathlib, dataclasses).
+- **No new pip dependencies**; pure stdlib (ast, pathlib, dataclasses, re).
 - **Read-only on `src/`**: the audit doesn't modify existing code. The new files are `src/code_path_audit.py` + `tests/test_code_path_audit.py` + the report under `docs/reports/code_path_audit/2026-06-07/`.
 - **Reusable**: re-run after any `src/` change to see drift. The 3 actions are declared once in `ACTIONS`; adding a 4th is one `Action(...)` declaration.
 - **Calibration target**: `pipeline_runtime_profiling_20260607` will use the existing `src/performance_monitor.py` to measure real costs and recalibrate `EXPENSIVE_THRESHOLD` + `COST_CLASS_WEIGHTS`.
@@ -9,7 +9,7 @@

 ## Overview

-Build `src/code_path_audit.py` — a data-oriented static-analysis tool that audits the 3 major actions (AI message lifecycle, discussion save/load, GUI startup) for expensive operations, redundant calls, and pipelining candidates. The output (JSON + markdown + Mermaid) is the artifact that informs pipeline-pruning decisions; the actual code changes are a follow-up track (`pipeline_pruning_20260607`).
+Build `src/code_path_audit.py` — a data-oriented static-analysis tool that audits the 3 major actions (AI message lifecycle, discussion save/load, GUI startup) for expensive operations, redundant calls, and pipelining candidates. The output (custom postfix `.dsl` data + markdown + Mermaid + prefix tree text) is the artifact that informs pipeline-pruning decisions; the actual code changes are a follow-up track (`pipeline_pruning_20260607`).

 Per the user's framing: "anything that can even remotely smell as an expensive bulk action or major action that takes more than 10-40 microseconds." The audit focuses on **expensive** operations (file I/O, network, AST parsing, big loops, anything that smells like a bulk action) inside the 3 actions — not on every state mutation. The cost model is heuristic, calibrated by a runtime-profiling follow-up (`pipeline_runtime_profiling_20260607`) that catches the cases static analysis can't resolve (C-extension cost, import cost, JIT effects, decorator-driven dispatch).

@@ -34,13 +34,13 @@ The MMA worker spawn action is **out of scope** for this track (per user: "keepi
 - A state-mutation index per function (5 mutation kinds: `attr_write`, `container_mutate`, `file_write`, `ipc_emit`, `global_write`).
 - An expensive-ops index (7 cost classes, with a heuristic data-size estimate).
 - A per-action traversal API (`trace_action(action, max_depth=10) -> ActionProfile`).
- An output suite: JSON data files + markdown summaries + Mermaid per-action call graphs.
+- An output suite: custom postfix `.dsl` data files + markdown summaries + Mermaid per-action call graphs + prefix-tree text view.
 - A CLI (`python -m src.code_path_audit --action <name>`) and an MCP tool (`code_path_audit(action_name, max_depth)`).
 - The actual audit run on the 3 actions, with the report committed to `docs/reports/code_path_audit/2026-06-07/`.

 ## Goals

-1. **Produce a queryable artifact.** The JSON output is the source of truth; markdown + Mermaid are for human review. Re-run after any `src/` change to see drift.
+1. **Produce a queryable artifact.** The custom postfix `.dsl` output is the source of truth; markdown + Mermaid + prefix-tree text are for human review. Re-run after any `src/` change to see drift.
 2. **Surface the top-N optimization candidates per action.** The `summary.md` ranks candidates by potential data-transform load reduction. This is what the user will use to decide which pruning/optimization work to do next.
 3. **Data-grounded design.** The audit's data structure is the spec; the heuristics and the threshold are module-level constants tunable from one place.
 4. **Reusable across actions.** The `trace_action` API takes any `Action` (entry point + description). Adding a 4th action (e.g., MMA worker spawn, when it's no longer cold) is one `Action(...)` declaration.
@@ -154,7 +154,7 @@ The user can extend with more actions later (e.g., MMA worker spawn when it's no

 CLI:
 ```bash
-uv run python -m src.code_path_audit --action ai_message_lifecycle [--depth N] [--json] [--markdown] [--mermaid]
+uv run python -m src.code_path_audit --action ai_message_lifecycle [--depth N] [--dsl] [--tree] [--markdown] [--mermaid]
 ```

 MCP tool (for agents):
@@ -166,16 +166,43 @@ Generated artifacts (all under `docs/reports/code_path_audit/<YYYY-MM-DD>/`):

 | File | Format | Purpose |
 |------|--------|---------|
-| `call_graph.json` | JSON | Full call graph (all of `src/`) |
-| `expensive_ops.json` | JSON | Expensive ops index (per-file, per-function) |
-| `state_mutations.json` | JSON | State mutations index (per function) |
-| `actions/<action>.json` | JSON | Per-action profile (machine-readable) |
-| `actions/<action>.md` | Markdown | Per-action human-readable summary |
+| `call_graph.dsl` | Custom postfix DSL | Full call graph (all of `src/`); machine-readable, parses in ~30 lines |
+| `expensive_ops.dsl` | Custom postfix DSL | Expensive ops index (per-file, per-function) |
+| `state_mutations.dsl` | Custom postfix DSL | State mutations index (per function) |
+| `actions/<action>.dsl` | Custom postfix DSL | Per-action profile (machine-readable) |
+| `actions/<action>.tree` | Prefix tree (text) | Per-action human-readable tree (for human review) |
+| `actions/<action>.md` | Markdown | Per-action summary + table (for code review) |
 | `actions/<action>.mmd` | Mermaid | Per-action call graph (visual) |
 | `summary.md` | Markdown | Top-level cross-action summary + ranked optimization candidates |
 | `optimization_candidates.md` | Markdown | Ranked list with: candidate, current cost, proposed reduction, effort, priority |

-The two follow-up tracks consume the JSON files; the markdown is for human review.
+The two follow-up tracks consume the .dsl files; the markdown + tree are for human review.
+
+**The custom DSL is postfix (RPN) with length-prefixed lists** — no brackets, no braces, no commas, no colons. Each "word" is a tagged constructor that consumes a known number of args from the stack (e.g., `fn` consumes 3, `exp-op` consumes 5, `mut` consumes 3, `N list` consumes N items). Whitespace-tokenized. Strings are bare atoms when they have no whitespace; quoted only when needed. `nil` for null. `\` for line comments. The DSL is deliberately NOT strict Forth — it's a custom postfix format tailored to the audit's record shapes (function, call, mutation, expensive op, pair, list).
+
+Example of a single FunctionNode record:
+
+```text
+\ FunctionNode: fqname file line fn
+"src.ai_client.AIClient.send" "src/ai_client.py" 100 fn
+"build_file_items" call
+"process_response" call
+"self.history" attr_write 110 mut
+"open" file_io 100 120 exp-op
+```
+
+**The prefix tree renderer** is a separate human-readable view of the same data — top-down, `├─`/`└─`/`│` box-drawing, scannable. Generated by a recursive walker. Inlined in the markdown reports (optionally produced as `actions/<action>.tree` for tooling).
+
+**Why custom postfix DSL (not JSON, not s-expressions, not strict Forth):**
+- **Not JSON** (JSON is ill-performant: quoting, escaping, hash table allocation, no streaming).
+- **Not s-expressions** (the bracket version drifts back toward s-exprs; the user wanted postfix specifically).
+- **Not strict Forth** (the user wants a format ideal for call-graph recording, not a Turing-complete Forth program).
+- **Postfix** (per user: "I want a post-fix heiarchy"): stack-based, no delimiters to count.
+- **Length-prefixed lists** (standard postfix solution for nesting): `N list` consumes N items, unambiguous.
+- **Trivial parser** (~30 lines: split + walk + evaluate tagged words against a known arity table).
+- **Compact**: ~30-40% fewer characters than JSON for the same data.
+- **Streamable**: no need to parse the whole file to find a record; you can scan for tags.
+- **Extensible**: add new metric types by adding new tagged words (`metric(name value sample_size)`, `histogram(buckets)`, etc.).

 ## Verification (TDD per `conductor/workflow.md`)

@@ -185,7 +212,8 @@ Unit tests in `tests/test_code_path_audit.py`:
 - `ExpensiveOpIndex` detects each of the 7 cost classes on synthetic source.
 - `StateMutationIndex` detects each of the 5 mutation kinds on synthetic source.
 - `trace_action` produces an `ActionProfile` for a synthetic action whose expected cost is computable by hand.
- JSON output round-trips (deserialize → same structure).
+- Custom postfix `.dsl` output round-trips (parse_dsl(to_dsl(profile)) == in-memory structure).
+- Prefix tree renderer produces well-formed box-drawing output for the 3 per-action reports.
 - Markdown output is well-formed (header per section, table per category).
 - Mermaid output parses as valid Mermaid syntax.

@@ -202,7 +230,7 @@ Manual verification: the report is the deliverable. A Tier 2 Tech Lead + user re
 2. feat(audit): add trace_action + ActionProfile + cost model
   - src/code_path_audit.py (extends with action tracing)
   - tests/test_code_path_audit.py (integration tests)
-3. feat(audit): add JSON / markdown / Mermaid output
+3. feat(audit): add custom postfix DSL writer + parser + tree renderer / markdown / Mermaid output
 4. feat(audit): add MCP tool + CLI surface
 5. docs(audit): run audit on 3 actions; commit report
   - docs/reports/code_path_audit/2026-06-07/* (the deliverable)
@@ -218,7 +246,7 @@ Each commit message includes a `git notes add -m "..."` summary per `conductor/w
 |------|-----------|--------|------------|
 | Heuristic cost model is imprecise; reported "expensive" ops aren't actually expensive at runtime. | Medium | Medium (false positives dilute the report) | `EXPENSIVE_THRESHOLD` is a module-level constant; the runtime-profiling follow-up calibrates it. |
 | AST walking misses dynamic patterns (eval, getattr, decorator-driven dispatch). | Medium | Medium (under-estimates some calls) | Document the limitations in the report's caveats section; the runtime-profiling follow-up catches these. |
-| Mermaid diagrams exceed renderable size for deep actions. | Medium | Low (visualization only) | Default `max_depth=5` for `--mermaid`; full graph available as JSON. |
+| Mermaid diagrams exceed renderable size for deep actions. | Medium | Low (visualization only) | Default `max_depth=5` for `--mermaid`; full graph available as `.dsl`. |
 | The 3 actions' entry points are not exactly the functions the user has in mind. | Medium | Low (the report is the artifact; user can re-run with different entry points) | Document the chosen entry points in the report; CLI/MCP tool accepts any fully-qualified function name. |
 | Report is too large to review (thousands of expensive ops). | Low | Medium | Per-action scoping; default `--depth 5`; ranked optimization candidates in `summary.md` make the top-N obvious. |
 | Existing `derive_code_path` is the de-facto call-graph tool and the new one is redundant. | Low | Low (the new one is a strict superset) | `derive_code_path` stays as a thin wrapper around `code_path_audit.trace_action` for backward compat, OR gets a `@deprecated` shim. |