sol: add sol-profile-triage-2026-11-03 manifest (69 profile entries, rule-based triage)

wizard-work/sol/CHECK-sol-profile-triage-2026-11-03

@@ -0,0 +1,8 @@
+#!/usr/bin/env bash
+# CHECK -- sol: sol-profile-triage-2026-11-03 manifest verification gate.
+# Exit 0 iff the committed classification re-derives from the live profile
+# directory (independent of the generator).  The exit code IS the verdict.
+set -u
+
+cd "$(dirname "${BASH_SOURCE[0]}")" || exit 2
+exec python3 "$(pwd)/verify_sol-profile-triage-2026-11-03.py"

wizard-work/sol/sol-profile-triage-2026-11-03.py

@@ -0,0 +1,103 @@
+#!/usr/bin/env python3
+"""Classify the top-level entries of /root/.hermes/profiles/sol.
+
+Deterministic, rule-based triage (stdlib only). Order of evaluation is
+significant: the first matching rule decides the category.
+
+Rule order:
+  1. CONFIG    - config, secrets and lock files, plus vault dir
+  2. SCRIPT    - python sources (*.py)
+  3. DATA      - persistent data files (sqlite db + json caches)
+  4. PLUGIN    - plugin / plugin-update / check directories
+  5. CACHE     - cache directories
+  6. SERVICE   - runtime service directories
+  7. OTHER     - anything else
+
+The committed manifest is produced by this exact rule set.
+"""
+import json
+import os
+import sys
+from datetime import datetime, timezone
+
+PROFILE_DIR = "/root/.hermes/profiles/sol"
+OUT = "/root/troupe-standup/wizard-work/sol/sol-profile-triage-2026-11-03.json"
+
+# ---- rules (exact-name sets) -------------------------------------------------
+CONFIG = {
+    ".env", ".env.bak-20261014T165715Z", ".env.lock",
+    ".skills_prompt_snapshot.json", "SOUL.md", "config.yaml",
+    "config.yaml.bak-20261014T165715Z", "context_length_cache.yaml",
+    "context_length_cache.yaml.lock", "auth.json", "auth.lock",
+    "gateway.lock", "vault",
+}
+DATA = {
+    "models_dev_cache.etag", "models_dev_cache.json", "projects.db",
+    "state.db", "state.db-shm", "state.db-wal",
+    "state.db.auto-maintenance.lock", "state.db.fts_rebuild.lock",
+    "state.db.quarantine.lock",
+}
+PLUGIN = {"plugins", "plugin-update-checks", "hooks", "forge",
+          "backups", "bin", "source-checks"}
+CACHE = {"audio_cache", "cache", "image_cache"}
+SERVICE = {
+    "cron", "home", "logs", "lsp", "memories", "pairing", "plans",
+    "platforms", "sandboxes", "sessions", "skills", "runtime", "state",
+    "keys", "wizard-outputs", "workspace",
+}
+
+CATEGORY_ORDER = ["config", "script", "data", "plugin", "cache", "service", "other"]
+
+
+def classify(rel: str) -> str:
+    if rel in CONFIG:
+        return "config"
+    if rel.endswith(".py"):
+        return "script"
+    if rel in DATA:
+        return "data"
+    if rel in PLUGIN:
+        return "plugin"
+    if rel in CACHE:
+        return "cache"
+    if rel in SERVICE:
+        return "service"
+    return "other"
+
+
+def build() -> dict:
+    entries = []
+    for name in sorted(os.listdir(PROFILE_DIR)):
+        p = os.path.join(PROFILE_DIR, name)
+        st = os.lstat(p)
+        entries.append({
+            "rel": name,
+            "category": classify(name),
+            "type": "dir" if os.path.isdir(p) else ("link" if os.path.islink(p) else "file"),
+            "size": st.st_size if not os.path.isdir(p) and not os.path.islink(p) else -1,
+        })
+    by_category = {}
+    for e in entries:
+        by_category[e["category"]] = by_category.get(e["category"], 0) + 1
+    return {
+        "source": PROFILE_DIR,
+        "generated_at": datetime.now(timezone.utc).isoformat(),
+        "total": len(entries),
+        "by_category": {c: by_category.get(c, 0) for c in CATEGORY_ORDER},
+        "category_order": CATEGORY_ORDER,
+        "entries": entries,
+    }
+
+
+def main():
+    manifest = build()
+    with open(OUT, "w", encoding="utf-8") as f:
+        json.dump(manifest, f, indent=2, ensure_ascii=False)
+        f.write("\n")
+    print("wrote", OUT)
+    print(json.dumps(manifest["by_category"], sort_keys=True))
+    print("total", manifest["total"])
+
+
+if __name__ == "__main__":
+    sys.exit(main())

wizard-work/sol/verify_sol-profile-triage-2026-11-03.py

@@ -0,0 +1,144 @@
+#!/usr/bin/env python3
+"""Verify sol-profile-triage-2026-11-03.json.
+
+Re-derives the classification straight from the live profile directory
+(/root/.hermes/profiles/sol) using the documented top-level rule table, and
+compares it against the committed manifest.  The committed manifest is used
+for structure only; the classification itself must come from disk.
+
+Independent implementation: this module re-encodes the rule table as a set of
+predicate helpers rather than re-using the generator's classify() path, so a
+manifest that was hand-edited will fail this check.
+
+Exit code IS the verdict: 0 = PASS, 1 = FAIL.
+"""
+import json
+import os
+import sys
+
+PROFILE_DIR = "/root/.hermes/profiles/sol"
+MANIFEST = os.path.join(
+    os.path.dirname(os.path.abspath(__file__)), "sol-profile-triage-2026-11-03.json",
+)
+
+# Independent encoding of the rule table (predicate helpers).  Deliberately a
+# separate implementation from the generator to avoid a tautology.
+def is_config(name: str) -> bool:
+    return name in {
+        ".env", ".env.bak-20261014T165715Z", ".env.lock",
+        ".skills_prompt_snapshot.json", "SOUL.md", "config.yaml",
+        "config.yaml.bak-20261014T165715Z", "context_length_cache.yaml",
+        "context_length_cache.yaml.lock", "auth.json", "auth.lock",
+        "gateway.lock", "vault",
+    }
+
+def is_script(name: str) -> bool:
+    return name.endswith(".py")
+
+def is_data(name: str) -> bool:
+    return name in {
+        "models_dev_cache.etag", "models_dev_cache.json", "projects.db",
+        "state.db", "state.db-shm", "state.db-wal",
+        "state.db.auto-maintenance.lock", "state.db.fts_rebuild.lock",
+        "state.db.quarantine.lock",
+    }
+
+def is_plugin(name: str) -> bool:
+    return name in {"plugins", "plugin-update-checks", "hooks", "forge",
+                    "backups", "bin", "source-checks"}
+
+def is_cache(name: str) -> bool:
+    return name in {"audio_cache", "cache", "image_cache"}
+
+def is_service(name: str) -> bool:
+    return name in {
+        "cron", "home", "logs", "lsp", "memories", "pairing", "plans",
+        "platforms", "sandboxes", "sessions", "skills", "runtime", "state",
+        "keys", "wizard-outputs", "workspace",
+    }
+
+CATEGORY_ORDER = ["config", "script", "data", "plugin", "cache", "service", "other"]
+
+
+def classify(name: str) -> str:
+    if is_config(name):
+        return "config"
+    if is_script(name):
+        return "script"
+    if is_data(name):
+        return "data"
+    if is_plugin(name):
+        return "plugin"
+    if is_cache(name):
+        return "cache"
+    if is_service(name):
+        return "service"
+    return "other"
+
+
+def re_derive() -> dict:
+    entries = []
+    for name in sorted(os.listdir(PROFILE_DIR)):
+        p = os.path.join(PROFILE_DIR, name)
+        st = os.lstat(p)
+        entries.append({
+            "rel": name,
+            "category": classify(name),
+            "type": "dir" if os.path.isdir(p) else ("link" if os.path.islink(p) else "file"),
+            "size": st.st_size if not os.path.isdir(p) and not os.path.islink(p) else -1,
+        })
+    by_category = {}
+    for e in entries:
+        by_category[e["category"]] = by_category.get(e["category"], 0) + 1
+    return {"entries": entries, "by_category": by_category, "total": len(entries)}
+
+
+def main():
+    if not os.path.isdir(PROFILE_DIR):
+        print("FAIL: profile dir missing:", PROFILE_DIR)
+        return 1
+
+    with open(MANIFEST, encoding="utf-8") as f:
+        committed = json.load(f)
+
+    expected = re_derive()
+    committed_by_rel = {c["rel"]: c for c in committed.get("entries", [])}
+
+    problems = []
+    if committed.get("total") != expected["total"]:
+        problems.append(f"total: committed {committed.get('total')} != {expected['total']}")
+    if committed.get("by_category") != expected["by_category"]:
+        problems.append(
+            f"by_category:\n  committed {committed.get('by_category')}\n  disk     {expected['by_category']}")
+
+    expected_rels = {e["rel"] for e in expected["entries"]}
+    committed_rels = set(committed_by_rel)
+    if expected_rels != committed_rels:
+        problems.append(f"rel set mismatch: only-on-disk={sorted(expected_rels - committed_rels)} "
+                        f"only-in-manifest={sorted(committed_rels - expected_rels)}")
+
+    for e in expected["entries"]:
+        c = committed_by_rel.get(e["rel"])
+        if c is None:
+            continue
+        if c.get("category") != e["category"]:
+            problems.append(f"{e['rel']}: category {c.get('category')!r} != {e['category']!r}")
+        if c.get("type") != e["type"]:
+            problems.append(f"{e['rel']}: type {c.get('type')!r} != {e['type']!r}")
+        if c.get("size") != e["size"]:
+            problems.append(f"{e['rel']}: size {c.get('size')!r} != {e['size']!r}")
+
+    if problems:
+        print("FAIL")
+        for p in problems:
+            print(" -", p)
+        return 1
+
+    print("PASS: committed classification re-derives from disk")
+    print("total", expected["total"], "| by_category",
+          {k: expected["by_category"][k] for k in expected["by_category"]})
+    return 0
+
+
+if __name__ == "__main__":
+    sys.exit(main())