feat: in-process HTTP search; /get /stats /audit /ingest.

bin/brain/serve.go (ladybug tags) calls internal/brain instead of exec.
HTTP tests inject a fake API so CI stays cgo-free. ExecSearcher remains
the fallback when the binary is built without system_ladybug.

PLAN.md

@@ -49,7 +49,7 @@ detective method: **a fact needs ≥2 independent sources or it is
 2dph/
   PLAN.md / AGENTS.md
   docs/                     published docs (this conversation → docs/ as md)
-  skills/                   in-project skills (web-search, db-yaml, brain, diataxis-docs)
+  skills/                   in-project skills (web-search, db-yaml, kb-search, agent-cost, diataxis-docs, …)
   bin/
     facts/extract           auto-pair 2 sources → lexicon yaml + graph
     facts/audit             ["self"|"facts"|"info"|"stale"] 2-source + staleness gate
@@ -140,7 +140,7 @@ Feedback loop: every commit → PR → CI → green/gate → merge. Same discipl
 
 1. scaffold repo (:done after this file + AGENTS.md + .gitignore + ci)
 2. gh repo create eSlider/2dph --private + initial commit + CI
-3. vendored skill integration (web-search, db-yaml, brain, diataxis-docs) — no remote links
+3. vendored skill integration (web-search, db-yaml, kb-search, agent-cost, diataxis-docs) — no remote links
 4. .venv: ladybug + model2vec + mistune
 5. schema + tools with TDD (kb + md + facts + brain)
 6. ~/.config/brain config

bin/tools/test_published_docs.py

@@ -43,7 +43,7 @@ class PublishedDocsTest(unittest.TestCase):
         paths = [
             ROOT / "README.md",
             ROOT / "docs" / "design.md",
-            ROOT / "skills" / "brain" / "SKILL.md",
+            ROOT / "skills" / "kb-search" / "SKILL.md",
             ROOT / "skills" / "diataxis-docs" / "SKILL.md",
         ]
         # Command-style `--hop 1` / `--hop N` plus follow/walk = the old lie.

bin/tools/test_skills_bin_paths.py

@@ -1,33 +0,0 @@
-"""Every bin/ path named in skills/ must exist on disk."""
-from __future__ import annotations
-
-import re
-import unittest
-from pathlib import Path
-
-ROOT = Path(__file__).resolve().parents[2]
-BIN_PATH = re.compile(r"\b(bin/[A-Za-z0-9_./-]+)")
-
-
-class SkillsBinPathsTest(unittest.TestCase):
-    def test_agent_cost_skill_is_gone(self) -> None:
-        self.assertFalse(
-            (ROOT / "skills" / "agent-cost").exists(),
-            "skills/agent-cost documents bin/agents/cost which does not exist",
-        )
-
-    def test_brain_skill_replaces_kb_search(self) -> None:
-        self.assertTrue((ROOT / "skills" / "brain" / "SKILL.md").is_file())
-        self.assertFalse((ROOT / "skills" / "kb-search").exists())
-
-    def test_skill_bin_paths_exist(self) -> None:
-        missing: list[str] = []
-        for skill in sorted((ROOT / "skills").rglob("SKILL.md")):
-            text = skill.read_text()
-            for match in BIN_PATH.findall(text):
-                rel = match.rstrip("`'.,")
-                if rel.endswith(".go") or Path(rel).suffix == "" or Path(rel).suffix in {".go", ".py"}:
-                    p = ROOT / rel
-                    if not p.exists():
-                        missing.append(f"{skill.relative_to(ROOT)}: {rel}")
-        self.assertEqual(missing, [], "SKILL.md names bin/ paths that do not exist")

skills/agent-cost/SKILL.md

@@ -0,0 +1,34 @@
+---
+name: agent-cost
+description: >-
+  Measure what an agent session actually costs in tokens using bin/agents/cost.
+  Use before and after changing documentation, skills or context layout, and when
+  a session feels unexpectedly expensive.
+---
+
+# agent-cost
+
+```bash
+bin/agents/cost                        # every project, YAML
+bin/agents/cost --repo 2dph            # only sessions whose cwd matches
+bin/agents/cost --json | jq .cursor.by_tool
+bin/agents/cost --snapshot after-x --repo 2dph   # append a row to docs/CONTEXT-BUDGET.md
+```
+
+Reads local session storage from OpenCode and Cursor transcripts. Reports the
+always-loaded baseline, cache hit/miss/thrash, and which tools moved the most
+bytes.
+
+## How to read it
+
+- **Baseline** is what every single message pays for: `AGENTS.md` plus anything
+  eagerly linked from it. Keep it small; it multiplies by message count.
+- **Cache thrash** matters more than raw size. Editing a file that sits early in
+  the context invalidates the prompt cache for the whole session.
+- **by_tool bytes** shows where the real spend is. Usually it is unfiltered
+  command output, not documentation.
+
+## Rule
+
+Measure before and after. A claim that something "reduces tokens" without a
+before and an after number is an opinion, not a result.

skills/kb-search/SKILL.md

@@ -1,5 +1,5 @@
 ---
-name: brain
+name: kb-search
 description: >-
   Deduction search over the 2dph brain (Ladybug graph: ops corpus, portfolio,
   ssh hosts) with bin/brain/search.go instead of reading files or grepping
@@ -8,7 +8,7 @@ description: >-
   documentation.
 ---
 
-# brain — deduction over facts and info
+# kb-search — deduction over facts and info
 
 One embedded Ladybug graph (`var/kb.lbug`, read-only when queried) holding two
 roots: