memindex 1.4.4 → 1.4.6
This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
- package/.claude-plugin/marketplace.json +1 -1
- package/.claude-plugin/plugin.json +1 -1
- package/.codex-plugin/plugin.json +1 -1
- package/.qoder-plugin/plugin.json +1 -1
- package/README.md +9 -1
- package/dist/bug-report/index.js +94 -83
- package/dist/npx-cli/index.js +250 -249
- package/dist/pi-extension/TYPEBOX-PROVENANCE.json +4 -4
- package/dist/pi-extension/index.js +9 -9
- package/integrations/dsh/package.json +1 -1
- package/integrations/openclaw/openclaw.plugin.json +1 -5
- package/integrations/pi/TYPEBOX-PROVENANCE.json +4 -4
- package/package.json +5 -5
- package/plugin/.claude-plugin/plugin.json +1 -1
- package/plugin/.codex-plugin/plugin.json +1 -1
- package/plugin/.qoder-plugin/plugin.json +1 -1
- package/plugin/package.json +1 -1
- package/plugin/scripts/context-generator.cjs +62 -61
- package/plugin/scripts/context-generator.cjs.map +3 -3
- package/plugin/scripts/mcp-server.cjs +2 -2
- package/plugin/scripts/transcript-watcher.cjs +2 -2
- package/plugin/scripts/transcript-watcher.cjs.map +1 -1
- package/plugin/scripts/worker-service.cjs +677 -669
- package/plugin/scripts/worker-service.cjs.map +3 -3
- package/plugin/skills/distill/SKILL.md +22 -0
- package/plugin/skills/how-it-works/SKILL.md +1 -1
- package/plugin/skills/how-it-works/onboarding-explainer.md +1 -1
- package/plugin/skills/import-chats/SKILL.md +22 -0
- package/plugin/skills/import-jira/SKILL.md +22 -0
- package/plugin/skills/import-reviews/SKILL.md +22 -0
- package/plugin/skills/pin/SKILL.md +26 -0
- package/plugin/skills/savings/SKILL.md +22 -0
- package/integrations/openclaw/skills/agent-cost-report/CHECKSUMS.txt +0 -50
- package/integrations/openclaw/skills/agent-cost-report/SKILL.md +0 -241
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/__init__.py +0 -1
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/behavior.py +0 -303
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/classify.py +0 -83
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/costs.py +0 -119
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/devices.py +0 -88
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/evidence.py +0 -160
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/labels.py +0 -115
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/measure.py +0 -89
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/mistakes.py +0 -310
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/patterns.py +0 -391
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/period.py +0 -98
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/prices.py +0 -95
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/render.py +0 -586
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rollup.py +0 -355
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rules.json +0 -4
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rules.py +0 -76
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/sample.py +0 -45
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/snapshot.py +0 -53
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/sync.py +0 -79
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/transcripts.py +0 -126
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr/wins.py +0 -341
- package/integrations/openclaw/skills/agent-cost-report/scripts/acr.py +0 -219
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/_db.py +0 -49
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/_paths.py +0 -9
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/codex/sessions/2026/09/18/rollout-codex-1.jsonl +0 -4
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/make_report_fixtures.py +0 -46
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/prices_sample.json +0 -6
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-a/sess-aaaa.jsonl +0 -8
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-b/sess-bbbb.jsonl +0 -1
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-1day.json +0 -823
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-30day.json +0 -2213
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-7day.json +0 -1252
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-empty.json +0 -536
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-session.json +0 -662
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_behavior.py +0 -299
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_devices.py +0 -99
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_labels.py +0 -96
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_measure.py +0 -120
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_pdf_windows.py +0 -54
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_period.py +0 -137
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_prices.py +0 -126
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_render.py +0 -238
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_render_encoding.py +0 -70
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_rollup.py +0 -263
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_snapshot.py +0 -63
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_sync.py +0 -52
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_transcript_encoding.py +0 -109
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_transcripts.py +0 -151
- package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_wins.py +0 -222
- package/plugin/skills/agent-cost-report/CHECKSUMS.txt +0 -50
- package/plugin/skills/agent-cost-report/SKILL.md +0 -241
- package/plugin/skills/agent-cost-report/scripts/acr/__init__.py +0 -1
- package/plugin/skills/agent-cost-report/scripts/acr/behavior.py +0 -303
- package/plugin/skills/agent-cost-report/scripts/acr/classify.py +0 -83
- package/plugin/skills/agent-cost-report/scripts/acr/costs.py +0 -119
- package/plugin/skills/agent-cost-report/scripts/acr/devices.py +0 -88
- package/plugin/skills/agent-cost-report/scripts/acr/evidence.py +0 -160
- package/plugin/skills/agent-cost-report/scripts/acr/labels.py +0 -115
- package/plugin/skills/agent-cost-report/scripts/acr/measure.py +0 -89
- package/plugin/skills/agent-cost-report/scripts/acr/mistakes.py +0 -310
- package/plugin/skills/agent-cost-report/scripts/acr/patterns.py +0 -391
- package/plugin/skills/agent-cost-report/scripts/acr/period.py +0 -98
- package/plugin/skills/agent-cost-report/scripts/acr/prices.py +0 -95
- package/plugin/skills/agent-cost-report/scripts/acr/render.py +0 -586
- package/plugin/skills/agent-cost-report/scripts/acr/rollup.py +0 -355
- package/plugin/skills/agent-cost-report/scripts/acr/rules.json +0 -4
- package/plugin/skills/agent-cost-report/scripts/acr/rules.py +0 -76
- package/plugin/skills/agent-cost-report/scripts/acr/sample.py +0 -45
- package/plugin/skills/agent-cost-report/scripts/acr/snapshot.py +0 -53
- package/plugin/skills/agent-cost-report/scripts/acr/sync.py +0 -79
- package/plugin/skills/agent-cost-report/scripts/acr/transcripts.py +0 -126
- package/plugin/skills/agent-cost-report/scripts/acr/wins.py +0 -341
- package/plugin/skills/agent-cost-report/scripts/acr.py +0 -219
- package/plugin/skills/agent-cost-report/scripts/tests/_db.py +0 -49
- package/plugin/skills/agent-cost-report/scripts/tests/_paths.py +0 -9
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/codex/sessions/2026/09/18/rollout-codex-1.jsonl +0 -4
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/make_report_fixtures.py +0 -46
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/prices_sample.json +0 -6
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-a/sess-aaaa.jsonl +0 -8
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-b/sess-bbbb.jsonl +0 -1
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-1day.json +0 -823
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-30day.json +0 -2213
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-7day.json +0 -1252
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-empty.json +0 -536
- package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-session.json +0 -662
- package/plugin/skills/agent-cost-report/scripts/tests/test_behavior.py +0 -299
- package/plugin/skills/agent-cost-report/scripts/tests/test_devices.py +0 -99
- package/plugin/skills/agent-cost-report/scripts/tests/test_labels.py +0 -96
- package/plugin/skills/agent-cost-report/scripts/tests/test_measure.py +0 -120
- package/plugin/skills/agent-cost-report/scripts/tests/test_pdf_windows.py +0 -54
- package/plugin/skills/agent-cost-report/scripts/tests/test_period.py +0 -137
- package/plugin/skills/agent-cost-report/scripts/tests/test_prices.py +0 -126
- package/plugin/skills/agent-cost-report/scripts/tests/test_render.py +0 -238
- package/plugin/skills/agent-cost-report/scripts/tests/test_render_encoding.py +0 -70
- package/plugin/skills/agent-cost-report/scripts/tests/test_rollup.py +0 -263
- package/plugin/skills/agent-cost-report/scripts/tests/test_snapshot.py +0 -63
- package/plugin/skills/agent-cost-report/scripts/tests/test_sync.py +0 -52
- package/plugin/skills/agent-cost-report/scripts/tests/test_transcript_encoding.py +0 -109
- package/plugin/skills/agent-cost-report/scripts/tests/test_transcripts.py +0 -151
- package/plugin/skills/agent-cost-report/scripts/tests/test_wins.py +0 -222
- package/plugin/skills/babysit/SKILL.md +0 -87
- package/plugin/skills/design-is/SKILL.md +0 -312
- package/plugin/skills/do/SKILL.md +0 -45
- package/plugin/skills/learn-codebase/SKILL.md +0 -21
- package/plugin/skills/make-plan/SKILL.md +0 -67
- package/plugin/skills/oh-my-issues/SKILL.md +0 -226
- package/plugin/skills/pathfinder/SKILL.md +0 -111
- package/plugin/skills/screenpipe/SKILL.md +0 -78
- package/plugin/skills/screenpipe/screenpipe.mjs +0 -196
- package/plugin/skills/smart-explore/SKILL.md +0 -193
- package/plugin/skills/standup/SKILL.md +0 -142
- package/plugin/skills/standup/agent-brief.md +0 -47
- package/plugin/skills/standup/standup.mjs +0 -586
- package/plugin/skills/what-the/SKILL.md +0 -6
- package/plugin/skills/wowerpoint/SKILL.md +0 -205
|
@@ -1,310 +0,0 @@
|
|
|
1
|
-
"""Phase 2B assembly (plan 2B.2, 2B.5, 2B.9): run the behavior pass, price the same-session union
|
|
2
|
-
of wasted turns once (the low figure = the headline), keep the upper bound in Details only, and hand
|
|
3
|
-
`behavior`, `spend.mistakes_*`, `timeline.mistakes_by_day` and per-line-item fields to the report.
|
|
4
|
-
|
|
5
|
-
One source of truth: the ribbon's waste sliver, line-item `wasted_cost` and the mistakes line all read
|
|
6
|
-
the union set built here. There is no second waste calculation anywhere else.
|
|
7
|
-
"""
|
|
8
|
-
import collections
|
|
9
|
-
import datetime as dt
|
|
10
|
-
import hashlib
|
|
11
|
-
|
|
12
|
-
from . import behavior, costs, patterns, prices, rules
|
|
13
|
-
from . import classify as classify_mod
|
|
14
|
-
from .period import PT
|
|
15
|
-
|
|
16
|
-
UPPER_LABEL = "upper bound, likely 5–10× too high (behavior study)"
|
|
17
|
-
MISTAKES_BASIS = "same-session wasted turns, each turn once"
|
|
18
|
-
TILE_MAX = 4
|
|
19
|
-
|
|
20
|
-
|
|
21
|
-
def classify_pass(flags, sessions, pricer, opts):
|
|
22
|
-
"""2B.4: only candidates the heuristics could not settle (P1 asks with no boost). A 'yes' confirms
|
|
23
|
-
the flag with label_source classifier; 'no' drops it. Spend is reported separately, never added."""
|
|
24
|
-
cands = []; by_id = {}
|
|
25
|
-
for i, f in enumerate(flags):
|
|
26
|
-
if f["pattern"] == "P1_invented_gates" and f.get("unconfirmed"):
|
|
27
|
-
S = sessions.get(f["session"]); users = [u for u in (S["users"] if S else []) if u["text"] and not u["results"]]
|
|
28
|
-
before = next((u["text"] for u in reversed(users) if u["ts_ms"] < (f["ts_ms"] or 0)), "")
|
|
29
|
-
after = next((u["text"] for u in users if u["ts_ms"] > (f["ts_ms"] or 0)), "")
|
|
30
|
-
cid = f"c{i}"; by_id[cid] = f; cands.append(dict(id=cid, kind="P1_invented_gates", text=f["excerpt"], before=before, after=after))
|
|
31
|
-
model = opts.get("model") or classify_mod.DEFAULT_MODEL
|
|
32
|
-
try:
|
|
33
|
-
labels, summary = classify_mod.run(cands, pricer.rate(model), model=model, cap_usd=opts.get("cap_usd") or classify_mod.CAP_USD, call=opts.get("call") or classify_mod._call)
|
|
34
|
-
except classify_mod.ClassifierError as ex:
|
|
35
|
-
return dict(ran=False, model=model, spend_usd=None, cap_usd=opts.get("cap_usd") or classify_mod.CAP_USD, sampled_n=0, error=str(ex))
|
|
36
|
-
for cid, lab in labels.items():
|
|
37
|
-
f = by_id[cid]; f["label_source"] = "classifier"; f["classifier"] = lab
|
|
38
|
-
if lab["label"] == "yes": f.pop("unconfirmed", None); f["basis"] += "; classifier: invented gate"
|
|
39
|
-
return summary
|
|
40
|
-
|
|
41
|
-
|
|
42
|
-
def _day_list(s_ms, e_ms):
|
|
43
|
-
d = dt.datetime.fromtimestamp(s_ms / 1000, PT).date(); end = dt.datetime.fromtimestamp(e_ms / 1000, PT).date()
|
|
44
|
-
out = []
|
|
45
|
-
while d < end: out.append(d.isoformat()); d += dt.timedelta(days=1)
|
|
46
|
-
return out
|
|
47
|
-
|
|
48
|
-
|
|
49
|
-
def default_model(sessions, pricer):
|
|
50
|
-
"""P3 comparison basis: the most-used priced Claude model in the window, else the plan's fallback."""
|
|
51
|
-
c = collections.Counter()
|
|
52
|
-
for S in sessions.values():
|
|
53
|
-
for T in S["turns"].values():
|
|
54
|
-
if T["model"] and T["usage"]:
|
|
55
|
-
m = prices.norm(T["model"])
|
|
56
|
-
if m.startswith("anthropic/claude") and pricer.usable(m): c[m] += 1
|
|
57
|
-
return c.most_common(1)[0][0] if c else patterns.P3_FALLBACK_DEFAULT
|
|
58
|
-
|
|
59
|
-
|
|
60
|
-
def mistake_id(session, first_key):
|
|
61
|
-
return "MK-" + hashlib.sha1(f"{session}|{first_key}".encode()).hexdigest()[:10]
|
|
62
|
-
|
|
63
|
-
|
|
64
|
-
def add_usage_turns(sessions, usage_rows, src="codex"):
|
|
65
|
-
"""Codex rows (usage.json) have a model id and tokens but no readable text on the box; they join as
|
|
66
|
-
text-less turns so P3 (wrong model) and the union see every logged model (R3)."""
|
|
67
|
-
for i, r in enumerate(usage_rows):
|
|
68
|
-
if r.get("src") != src: continue
|
|
69
|
-
S = sessions.setdefault(r["session"], behavior.new_session()); S.setdefault("tag", "unknown")
|
|
70
|
-
key = (f"{src}:{r['session']}", i)
|
|
71
|
-
S["turns"][key] = dict(key=key, ts_ms=int(behavior.transcripts.ts(r["ts"]).timestamp() * 1000), model=r["model"], text="", tool_uses=[],
|
|
72
|
-
usage={f: r[f] for f in behavior.FIELDS}, sidechain=False, file=r.get("file"), uuid=None)
|
|
73
|
-
|
|
74
|
-
|
|
75
|
-
UNTAGGED = "unavailable (prompts not tagged human vs bot)"
|
|
76
|
-
|
|
77
|
-
|
|
78
|
-
ALTERNATES = ("P4_over_engineering", "P5_not_asked", "P8_wrong_tool", "P10_memory_loss")
|
|
79
|
-
PRECISION_GATE = 0.70
|
|
80
|
-
|
|
81
|
-
|
|
82
|
-
def precision_gate(pats, precision):
|
|
83
|
-
"""Phase 8.5 / 2B.5: a tile pattern whose spot-check precision is under 70% moves to Details as low
|
|
84
|
-
confidence; the next pattern by ranking takes its place only if it passes."""
|
|
85
|
-
if not precision: return
|
|
86
|
-
freed = 0
|
|
87
|
-
for p in pats:
|
|
88
|
-
pr = precision.get(p["key"])
|
|
89
|
-
if pr is None: continue
|
|
90
|
-
p["precision"] = pr
|
|
91
|
-
if p["placement"] == "tile" and pr < PRECISION_GATE: p["placement"] = "details"; p["confidence"] = "low"; p["gate"] = f"spot-check precision {pr:.0%} < 70%: moved to Details"; freed += 1
|
|
92
|
-
for key in ALTERNATES:
|
|
93
|
-
if freed <= 0: break
|
|
94
|
-
p = next((x for x in pats if x["key"] == key), None)
|
|
95
|
-
if p and precision.get(key) is not None and precision[key] >= PRECISION_GATE: p["placement"] = "tile"; p["gate"] = f"promoted: precision {precision[key]:.0%}"; freed -= 1
|
|
96
|
-
|
|
97
|
-
|
|
98
|
-
def run(db, scope, window_block, pricer, session_projects, items, classify=None, rules_dir=None, now_ms=None, usage_rows=(),
|
|
99
|
-
glob_pattern=None, tagger=True, precision=None):
|
|
100
|
-
"""session_projects: {content_session_id: project}. items: Phase 2 line items (mutated: behavior_counts,
|
|
101
|
-
wasted_cost, recovery_cost, productive_cost, failure_signals). tagger=False models the R2 gate: no count
|
|
102
|
-
that reads user messages runs, and those counts render UNTAGGED."""
|
|
103
|
-
s_ms, e_ms = scope.S, scope.E
|
|
104
|
-
sessions, scan_stats = behavior.scan(s_ms, e_ms, session=scope.session if scope.kind == "session" else None, **({"pattern": glob_pattern} if glob_pattern else {}))
|
|
105
|
-
if scope.kind == "project":
|
|
106
|
-
sessions = {sid: S for sid, S in sessions.items() if sid in session_projects}
|
|
107
|
-
for S in sessions.values(): behavior.tag_user_turns(S)
|
|
108
|
-
add_usage_turns(sessions, usage_rows)
|
|
109
|
-
behavior.price_turns(sessions, pricer)
|
|
110
|
-
if tagger:
|
|
111
|
-
prompt_rows = behavior.load_prompt_rows(db, scope)
|
|
112
|
-
if scope.kind == "project": prompt_rows = [r for r in prompt_rows if r["session"] in session_projects]
|
|
113
|
-
msgs, author_tags = behavior.human_messages(sessions, prompt_rows)
|
|
114
|
-
else:
|
|
115
|
-
prompt_rows = []; msgs = {}; author_tags = UNTAGGED
|
|
116
|
-
for S in sessions.values():
|
|
117
|
-
for u in S["users"]: u["author"] = "unknown" if u.get("author") != "tool" else "tool"
|
|
118
|
-
tagger_ran = tagger
|
|
119
|
-
dm = default_model(sessions, pricer)
|
|
120
|
-
|
|
121
|
-
flags = []; denials = []; stats = collections.Counter()
|
|
122
|
-
for sid, S in sessions.items():
|
|
123
|
-
f, d, st = patterns.run_session(S, sid, msgs, dm, pricer); flags += f; denials += d
|
|
124
|
-
for k, v in st.items(): stats[k] += v
|
|
125
|
-
turn_by_key = {T["key"]: T for S in sessions.values() for T in S["turns"].values()}
|
|
126
|
-
classifier_block = classify_pass(flags, sessions, pricer, classify) if classify and classify.get("enabled") else None
|
|
127
|
-
flagged = collections.defaultdict(set)
|
|
128
|
-
for f in flags:
|
|
129
|
-
if f.get("unconfirmed"): continue
|
|
130
|
-
if f["key"] is not None: flagged[f["pattern"]].add(f["key"])
|
|
131
|
-
for k in f.get("extra_keys", []) + f.get("wasted_extra", []): flagged[f["pattern"]].add(k)
|
|
132
|
-
flagged["S2_hedging"] = {f["key"] for f in flags if f["pattern"] == "S2_hedging" and f.get("priced")}
|
|
133
|
-
flagged["P11_jargon"] = set()
|
|
134
|
-
flagged["P9_bad_outbound"] = set()
|
|
135
|
-
redo_keys = collections.defaultdict(set)
|
|
136
|
-
for f in flags:
|
|
137
|
-
for k in f.get("recovery_keys", []): redo_keys["S1_tool_errors"].add(k)
|
|
138
|
-
|
|
139
|
-
episodes = []; ep_flag_keys = set(); ep_redo_keys = set(); project_fallback = set()
|
|
140
|
-
for ep in behavior.find_episodes(msgs, phrase_hit=patterns.human_phrase_patterns):
|
|
141
|
-
S = sessions.get(ep["session"])
|
|
142
|
-
wasted, redo, last_agent = behavior.episode_windows(ep, S, msgs)
|
|
143
|
-
wkeys = {T["key"] for T in wasted}
|
|
144
|
-
pat = next((p for p in patterns.TILE_ORDER + tuple(k for k, *_ in patterns.PATTERNS) if flagged.get(p) and flagged[p] & wkeys), None)
|
|
145
|
-
if pat is None:
|
|
146
|
-
hp = collections.Counter(p for m in ep["messages"] for p in patterns.human_phrase_patterns(m["text"]))
|
|
147
|
-
order = [k for k, *_ in patterns.PATTERNS]
|
|
148
|
-
pat = min(hp, key=lambda k: (-hp[k], order.index(k))) if hp else "P12_unclear"
|
|
149
|
-
measured = S is not None and bool(S["turns"])
|
|
150
|
-
if measured:
|
|
151
|
-
ep_flag_keys |= wkeys; ep_redo_keys |= {T["key"] for T in redo}
|
|
152
|
-
flagged[pat] |= wkeys
|
|
153
|
-
else:
|
|
154
|
-
proj = session_projects.get(ep["session"])
|
|
155
|
-
if proj:
|
|
156
|
-
for sid2, S2 in sessions.items():
|
|
157
|
-
if session_projects.get(sid2) == proj:
|
|
158
|
-
for T in S2["turns"].values():
|
|
159
|
-
if ep["first_ts"] - behavior.WASTED_CAP_MS <= T["ts_ms"] < ep["first_ts"]: project_fallback.add(T["key"])
|
|
160
|
-
model = None
|
|
161
|
-
if S:
|
|
162
|
-
mc = collections.Counter(prices.norm(T["model"]) for T in S["turns"].values() if T["model"])
|
|
163
|
-
model = mc.most_common(1)[0][0] if mc else None
|
|
164
|
-
low = behavior.turn_cost(wasted) if measured else None
|
|
165
|
-
high = (low + behavior.turn_cost(redo)) if measured else None
|
|
166
|
-
episodes.append(dict(episode_id=ep["episode_id"], session=ep["session"], ts_pt=behavior.pt_iso(ep["first_ts"]), day_pt=behavior.day_pt(ep["first_ts"]),
|
|
167
|
-
pattern=pat, messages_n=len(ep["messages"]), low_usd=costs.usd(low) if measured else None, high_usd=costs.usd(high) if measured else None,
|
|
168
|
-
cost_status="measured_tokens" if measured else "unmeasured", user_minutes=behavior.user_minutes(ep, last_agent),
|
|
169
|
-
model=model, model_status="observed" if model else "assumed", source=ep["messages"][0]["source"],
|
|
170
|
-
device=ep["messages"][0].get("device", "local"), excerpt=behavior.scrub(ep["messages"][0]["text"]), wasted_keys=[list(k) for k in wkeys]))
|
|
171
|
-
|
|
172
|
-
span_keys = collections.defaultdict(set)
|
|
173
|
-
for sid, S in sessions.items():
|
|
174
|
-
for ft, keys in behavior.failure_spans(S, msgs).items(): span_keys[ft] |= keys
|
|
175
|
-
for f in flags:
|
|
176
|
-
if f["pattern"] == "S1_tool_errors" and f.get("loop"): span_keys["Looping"].add(f["key"])
|
|
177
|
-
|
|
178
|
-
union = set().union(*flagged.values()) | ep_flag_keys | set().union(*span_keys.values()) if (flagged or span_keys) else set(ep_flag_keys)
|
|
179
|
-
union = {k for k in union if k in turn_by_key}
|
|
180
|
-
redo_all = (ep_redo_keys | set().union(*redo_keys.values()) if redo_keys else ep_redo_keys) - union
|
|
181
|
-
high_set = union | redo_all | (project_fallback - union)
|
|
182
|
-
low_x1e6 = sum(turn_by_key[k]["x1e6"] or 0 for k in union)
|
|
183
|
-
high_x1e6 = sum(turn_by_key[k]["x1e6"] or 0 for k in high_set)
|
|
184
|
-
unmeasured_n = sum(1 for e in episodes if e["cost_status"] == "unmeasured")
|
|
185
|
-
|
|
186
|
-
by_ep = collections.Counter(e["pattern"] for e in episodes)
|
|
187
|
-
ex_by_pat = collections.defaultdict(list)
|
|
188
|
-
for f in flags:
|
|
189
|
-
if not f.get("unconfirmed") and len(ex_by_pat[f["pattern"]]) < 10:
|
|
190
|
-
ex_by_pat[f["pattern"]].append(dict(content_session_id=f["session"], ts_pt=behavior.pt_iso(f["ts_ms"]) if f["ts_ms"] else None, excerpt=f["excerpt"], basis=f["basis"], label_source=f["label_source"]))
|
|
191
|
-
pats = []
|
|
192
|
-
for key, name, mi, ft, place in patterns.PATTERNS:
|
|
193
|
-
keys = flagged.get(key, set()); fl = [f for f in flags if f["pattern"] == key and not f.get("unconfirmed")]
|
|
194
|
-
count = len(fl) + by_ep.get(key, 0) if key not in ("P12_unclear",) else by_ep.get(key, 0)
|
|
195
|
-
low = None if key == "P9_bad_outbound" else costs.usd(sum(turn_by_key[k]["x1e6"] or 0 for k in keys if k in turn_by_key))
|
|
196
|
-
hk = keys | {k for k in redo_all if key == "S1_tool_errors"} | set().union(*[set(map(tuple, e["wasted_keys"])) for e in episodes if e["pattern"] == key] or [set()])
|
|
197
|
-
high = None if key == "P9_bad_outbound" else costs.usd(sum(turn_by_key[k]["x1e6"] or 0 for k in hk if k in turn_by_key))
|
|
198
|
-
p = dict(key=key, name=name, source_ref=f"metrics-ideas.md:{mi}" if mi else "plan 2B.3 supporting metric", failure_type=ft, placement=place if key in patterns.TILE_ORDER else "details",
|
|
199
|
-
count=count, count_basis="heuristic", low_usd=low, high_usd=high, high_label=UPPER_LABEL, unmeasured_n=sum(1 for e in episodes if e["pattern"] == key and e["cost_status"] == "unmeasured"),
|
|
200
|
-
user_minutes=round(sum(e["user_minutes"] for e in episodes if e["pattern"] == key) + sum(f.get("user_minutes", 0) for f in fl), 1),
|
|
201
|
-
agent_minutes=None, waiting_minutes=round(sum(f.get("waiting_minutes") or 0 for f in fl), 1) if key == "P1_invented_gates" else None,
|
|
202
|
-
confidence="high" if key in ("P3_wrong_model", "S1_tool_errors", "P9_bad_outbound") else "medium", overlaps=True, examples=ex_by_pat.get(key, []),
|
|
203
|
-
detection="H" if key != "P12_unclear" else "base detector; M attributes when it can")
|
|
204
|
-
if key == "P3_wrong_model":
|
|
205
|
-
p["delta_usd"] = costs.usd(sum(max(f.get("delta_x1e6") or 0, 0) for f in fl)); p["default_model"] = dm; p["model_not_logged"] = stats["model_not_logged"]
|
|
206
|
-
if key == "P1_invented_gates":
|
|
207
|
-
p["unconfirmed_n"] = sum(1 for f in flags if f["pattern"] == key and f.get("unconfirmed")); p["never_answered_n"] = sum(1 for f in fl if f.get("never_answered"))
|
|
208
|
-
if key == "P9_bad_outbound":
|
|
209
|
-
p["incidents"] = len(fl); p["recipients"] = sum(f.get("recipients") or 0 for f in fl); p["cost_note"] = "incidents × recipients, never dollars"
|
|
210
|
-
if key == "S1_tool_errors": p["error_chains"] = stats["error_chains"]; p["loops"] = stats["loops"]; p["recovered_usd"] = costs.usd(sum(turn_by_key[k]["x1e6"] or 0 for k in redo_keys["S1_tool_errors"] if k in turn_by_key))
|
|
211
|
-
if key == "S2_hedging": p["availability_unknown_n"] = stats["hedge_availability_unknown"]; p["cost_note"] = "text only, not priced" if not keys else "follow-up round trip priced as recovery"
|
|
212
|
-
if key == "P11_jargon": p["cost_note"] = "user-minutes = words ÷ 200; dollars only for a clarification round trip"
|
|
213
|
-
pats.append(p)
|
|
214
|
-
precision_gate(pats, precision)
|
|
215
|
-
tiles = [p for p in pats if p["placement"] == "tile"]
|
|
216
|
-
assert len(tiles) <= 5 and len({("tile2" if p["key"] in patterns.MERGED_TILE else p["key"]) for p in tiles}) <= TILE_MAX
|
|
217
|
-
|
|
218
|
-
per_session = collections.defaultdict(lambda: dict(low=0, redo=0, counts=collections.Counter(), failures=set()))
|
|
219
|
-
for k in union:
|
|
220
|
-
T = turn_by_key[k]; sid = next(s for s, S in sessions.items() if k in S["turns"]); per_session[sid]["low"] += T["x1e6"] or 0
|
|
221
|
-
for k in redo_all:
|
|
222
|
-
T = turn_by_key[k]; sid = next(s for s, S in sessions.items() if k in S["turns"]); per_session[sid]["redo"] += T["x1e6"] or 0
|
|
223
|
-
for f in flags:
|
|
224
|
-
if f.get("unconfirmed"): continue
|
|
225
|
-
per_session[f["session"]]["counts"][f["pattern"]] += 1
|
|
226
|
-
ft = f.get("failure_type") or dict((k, t) for k, _, _, t, _ in patterns.PATTERNS).get(f["pattern"])
|
|
227
|
-
if ft: per_session[f["session"]]["failures"].add(ft)
|
|
228
|
-
for e in episodes: per_session[e["session"]]["counts"][e["pattern"]] += 1
|
|
229
|
-
def _reconciled(field, total_x1e6):
|
|
230
|
-
"""Round per item, then give the cent remainder to the largest item so the items sum to the total exactly (8.6)."""
|
|
231
|
-
vals = {li["content_session_id"]: (per_session[li["content_session_id"]][field] if li["content_session_id"] in per_session else 0) for li in items}
|
|
232
|
-
rounded = {k: costs.usd(v) for k, v in vals.items()}; diff = round(costs.usd(total_x1e6) - sum(rounded.values()), 2)
|
|
233
|
-
if diff and vals:
|
|
234
|
-
k = max(vals, key=vals.get); rounded[k] = round(rounded[k] + diff, 2)
|
|
235
|
-
return rounded
|
|
236
|
-
in_items = {li["content_session_id"] for li in items}
|
|
237
|
-
low_in_items = sum(v["low"] for k, v in per_session.items() if k in in_items); redo_in_items = sum(v["redo"] for k, v in per_session.items() if k in in_items)
|
|
238
|
-
wasted_by = _reconciled("low", low_in_items); redo_by = _reconciled("redo", redo_in_items)
|
|
239
|
-
for li in items:
|
|
240
|
-
ps = per_session.get(li["content_session_id"])
|
|
241
|
-
li["behavior_counts"] = dict(ps["counts"]) if ps else {}
|
|
242
|
-
li["wasted_cost"] = wasted_by.get(li["content_session_id"], 0.0)
|
|
243
|
-
li["recovery_cost"] = redo_by.get(li["content_session_id"], 0.0)
|
|
244
|
-
li["productive_cost"] = round(max(li["attributed_usd"] - li["wasted_cost"], 0.0), 2)
|
|
245
|
-
if ps and ps["failures"]:
|
|
246
|
-
li["failure_signals"] = sorted(set(li["failure_signals"]) | ps["failures"]); li["failure_type"] = li["failure_type"] or li["failure_signals"][0]
|
|
247
|
-
|
|
248
|
-
days = _day_list(s_ms, e_ms) if scope.kind != "session" else sorted({behavior.day_pt(T["ts_ms"]) for T in turn_by_key.values()} | {e["day_pt"] for e in episodes})
|
|
249
|
-
by_day = {d: dict(day_pt=d, usd=0, high_usd=0, unmeasured_n=0, user_minutes=0.0, turns_n=0, by_pattern=collections.Counter(), outbound_incidents=0, top_examples=[]) for d in days}
|
|
250
|
-
key_pats = collections.defaultdict(set)
|
|
251
|
-
for p, ks in flagged.items():
|
|
252
|
-
for k in ks: key_pats[k].add(p)
|
|
253
|
-
for k in union:
|
|
254
|
-
T = turn_by_key[k]; d = by_day.get(behavior.day_pt(T["ts_ms"]))
|
|
255
|
-
if d is None: continue
|
|
256
|
-
d["usd"] += T["x1e6"] or 0; d["turns_n"] += 1
|
|
257
|
-
for p in key_pats.get(k, ()): d["by_pattern"][p] += T["x1e6"] or 0
|
|
258
|
-
for k in high_set:
|
|
259
|
-
T = turn_by_key[k]; d = by_day.get(behavior.day_pt(T["ts_ms"]))
|
|
260
|
-
if d is not None: d["high_usd"] += T["x1e6"] or 0
|
|
261
|
-
for e in episodes:
|
|
262
|
-
d = by_day.get(e["day_pt"])
|
|
263
|
-
if d is None: continue
|
|
264
|
-
d["user_minutes"] += e["user_minutes"]
|
|
265
|
-
if e["cost_status"] == "unmeasured": d["unmeasured_n"] += 1
|
|
266
|
-
if len(d["top_examples"]) < 3: d["top_examples"].append(dict(mistake_id=mistake_id(e["session"], e["episode_id"]), episode_id=e["episode_id"], pattern=e["pattern"], content_session_id=e["session"], ts_pt=e["ts_pt"]))
|
|
267
|
-
for f in flags:
|
|
268
|
-
if f["pattern"] == "P9_bad_outbound" and f["ts_ms"] and behavior.day_pt(f["ts_ms"]) in by_day: by_day[behavior.day_pt(f["ts_ms"])]["outbound_incidents"] += 1
|
|
269
|
-
for d in by_day.values():
|
|
270
|
-
d["usd"] = costs.usd(d["usd"]); d["high_usd"] = costs.usd(d["high_usd"]); d["user_minutes"] = round(d["user_minutes"], 1)
|
|
271
|
-
d["by_pattern"] = {p: costs.usd(v) for p, v in d["by_pattern"].items()}
|
|
272
|
-
clusters = []
|
|
273
|
-
for sid, S in sessions.items():
|
|
274
|
-
run = []
|
|
275
|
-
for T in behavior.ordered_turns(S):
|
|
276
|
-
if T["key"] in union: run.append(T)
|
|
277
|
-
elif run: clusters.append((sid, run)); run = []
|
|
278
|
-
if run: clusters.append((sid, run))
|
|
279
|
-
mistake_ids = [dict(mistake_id=mistake_id(sid, r[0]["key"]), content_session_id=sid, ts_pt=behavior.pt_iso(r[0]["ts_ms"]), turns_n=len(r),
|
|
280
|
-
usd=costs.usd(sum(T["x1e6"] or 0 for T in r)), patterns=sorted(set().union(*[key_pats.get(T["key"], set()) for T in r]))) for sid, r in clusters]
|
|
281
|
-
|
|
282
|
-
rule_rows = rules.effectiveness(episodes, msgs, s_ms, e_ms, rules_dir=rules_dir)
|
|
283
|
-
|
|
284
|
-
outbound = dict(incidents=sum(1 for f in flags if f["pattern"] == "P9_bad_outbound"), recipients=sum(f.get("recipients") or 0 for f in flags if f["pattern"] == "P9_bad_outbound"), user_minutes=None)
|
|
285
|
-
user_known = [e["user_minutes"] for e in episodes if e["user_minutes"] is not None]
|
|
286
|
-
unmatched_x1e6 = sum(turn_by_key[k]["x1e6"] or 0 for k in union if next(s for s, S in sessions.items() if k in S["turns"]) not in session_projects)
|
|
287
|
-
spend = dict(mistakes_estimated_usd=costs.usd(low_x1e6), mistakes_label="ESTIMATED", mistakes_basis=MISTAKES_BASIS, mistakes_turns_n=len(union),
|
|
288
|
-
mistakes_unmatched_usd=costs.usd(unmatched_x1e6),
|
|
289
|
-
mistakes_episodes_n=len(episodes), mistakes_unmeasured_n=unmeasured_n, mistakes_user_minutes=round(sum(user_known), 1), mistakes_user_minutes_known_n=len(user_known),
|
|
290
|
-
mistakes_high_usd=costs.usd(high_x1e6), mistakes_high_label=UPPER_LABEL)
|
|
291
|
-
block = dict(patterns=pats, outbound=outbound, episodes=[{k: v for k, v in e.items() if k != "wasted_keys"} for e in episodes],
|
|
292
|
-
union_low_usd=costs.usd(low_x1e6), union_high_usd=costs.usd(high_x1e6), union_turns_n=len(union), high_turns_n=len(high_set),
|
|
293
|
-
author_tags=author_tags, tagger_ran=tagger_ran, session_tags=collections.Counter(S.get("tag", "unknown") for S in sessions.values()),
|
|
294
|
-
human_message_counts=("computed from human-tagged turns" if tagger else UNTAGGED),
|
|
295
|
-
episodes_status=("measured" if tagger else UNTAGGED),
|
|
296
|
-
rule_effectiveness=rule_rows, coverage=dict(box="measured", mac="unavailable (no Phase 5 export)", grok_bot="unavailable"),
|
|
297
|
-
classifier=classifier_block or dict(ran=False, model=None, spend_usd=None, cap_usd=classify_mod.CAP_USD, sampled_n=0, note="off by default (G8); enable with --classify"),
|
|
298
|
-
permission_denials=dict(count=len(denials), examples=denials[:10], note="blocked by a rule; not priced as agent waste"),
|
|
299
|
-
label_sources=collections.Counter(f["label_source"] for f in flags if not f.get("unconfirmed")),
|
|
300
|
-
unconfirmed=dict(P1_candidates=[dict(content_session_id=f["session"], ts_pt=behavior.pt_iso(f["ts_ms"]), excerpt=f["excerpt"]) for f in flags if f.get("unconfirmed")][:50]),
|
|
301
|
-
default_model=dm, scan=scan_stats, mistake_clusters=mistake_ids, prompt_rows_n=len(prompt_rows),
|
|
302
|
-
not_in_mistakes_line="session-level keyword failure signals with no turn span stay in Details (2B.9)")
|
|
303
|
-
raw = dict(flags=[dict(pattern=f["pattern"], session=f["session"], ts_pt=behavior.pt_iso(f["ts_ms"]) if f.get("ts_ms") else None, excerpt=f["excerpt"], basis=f["basis"],
|
|
304
|
-
label_source=f["label_source"], unconfirmed=bool(f.get("unconfirmed"))) for f in flags],
|
|
305
|
-
user_turns=[dict(session=sid, ts_pt=behavior.pt_iso(m["ts_ms"]), author=m["author"], source=m["source"], excerpt=behavior.scrub(m["text"])) for sid, ms in msgs.items() for m in ms],
|
|
306
|
-
episodes=block["episodes"],
|
|
307
|
-
sessions=[dict(session=sid, tag=S.get("tag", "unknown"), turns=len(S["turns"]), tool_errors=sum(1 for u in S["users"] for r in u["results"] if r["is_error"]),
|
|
308
|
-
last_text=behavior.scrub(next((behavior.user_facing(T["text"]) for T in reversed(behavior.ordered_turns(S)) if T["text"].strip()), ""))) for sid, S in sessions.items()])
|
|
309
|
-
block["_raw"] = raw
|
|
310
|
-
return block, spend, list(by_day.values())
|