memindex 1.4.5 → 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.
Files changed (149) hide show
  1. package/.claude-plugin/marketplace.json +1 -1
  2. package/.claude-plugin/plugin.json +1 -1
  3. package/.codex-plugin/plugin.json +1 -1
  4. package/.qoder-plugin/plugin.json +1 -1
  5. package/README.md +7 -1
  6. package/dist/bug-report/index.js +94 -83
  7. package/dist/npx-cli/index.js +3 -3
  8. package/dist/pi-extension/TYPEBOX-PROVENANCE.json +4 -4
  9. package/dist/pi-extension/index.js +9 -9
  10. package/integrations/dsh/package.json +1 -1
  11. package/integrations/openclaw/openclaw.plugin.json +1 -5
  12. package/integrations/pi/TYPEBOX-PROVENANCE.json +4 -4
  13. package/package.json +5 -5
  14. package/plugin/.claude-plugin/plugin.json +1 -1
  15. package/plugin/.codex-plugin/plugin.json +1 -1
  16. package/plugin/.qoder-plugin/plugin.json +1 -1
  17. package/plugin/package.json +1 -1
  18. package/plugin/scripts/context-generator.cjs +62 -61
  19. package/plugin/scripts/context-generator.cjs.map +3 -3
  20. package/plugin/scripts/mcp-server.cjs +2 -2
  21. package/plugin/scripts/transcript-watcher.cjs +2 -2
  22. package/plugin/scripts/transcript-watcher.cjs.map +1 -1
  23. package/plugin/scripts/worker-service.cjs +677 -669
  24. package/plugin/scripts/worker-service.cjs.map +3 -3
  25. package/plugin/skills/distill/SKILL.md +22 -0
  26. package/plugin/skills/how-it-works/SKILL.md +1 -1
  27. package/plugin/skills/how-it-works/onboarding-explainer.md +1 -1
  28. package/plugin/skills/import-chats/SKILL.md +22 -0
  29. package/plugin/skills/import-jira/SKILL.md +22 -0
  30. package/plugin/skills/import-reviews/SKILL.md +22 -0
  31. package/plugin/skills/pin/SKILL.md +26 -0
  32. package/plugin/skills/savings/SKILL.md +22 -0
  33. package/integrations/openclaw/skills/agent-cost-report/CHECKSUMS.txt +0 -50
  34. package/integrations/openclaw/skills/agent-cost-report/SKILL.md +0 -241
  35. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/__init__.py +0 -1
  36. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/behavior.py +0 -303
  37. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/classify.py +0 -83
  38. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/costs.py +0 -119
  39. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/devices.py +0 -88
  40. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/evidence.py +0 -160
  41. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/labels.py +0 -115
  42. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/measure.py +0 -89
  43. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/mistakes.py +0 -310
  44. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/patterns.py +0 -391
  45. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/period.py +0 -98
  46. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/prices.py +0 -95
  47. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/render.py +0 -586
  48. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rollup.py +0 -355
  49. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rules.json +0 -4
  50. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/rules.py +0 -76
  51. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/sample.py +0 -45
  52. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/snapshot.py +0 -53
  53. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/sync.py +0 -79
  54. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/transcripts.py +0 -126
  55. package/integrations/openclaw/skills/agent-cost-report/scripts/acr/wins.py +0 -341
  56. package/integrations/openclaw/skills/agent-cost-report/scripts/acr.py +0 -219
  57. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/_db.py +0 -49
  58. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/_paths.py +0 -9
  59. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/codex/sessions/2026/09/18/rollout-codex-1.jsonl +0 -4
  60. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/make_report_fixtures.py +0 -46
  61. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/prices_sample.json +0 -6
  62. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-a/sess-aaaa.jsonl +0 -8
  63. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-b/sess-bbbb.jsonl +0 -1
  64. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-1day.json +0 -823
  65. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-30day.json +0 -2213
  66. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-7day.json +0 -1252
  67. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-empty.json +0 -536
  68. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/fixtures/report-session.json +0 -662
  69. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_behavior.py +0 -299
  70. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_devices.py +0 -99
  71. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_labels.py +0 -96
  72. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_measure.py +0 -120
  73. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_pdf_windows.py +0 -54
  74. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_period.py +0 -137
  75. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_prices.py +0 -126
  76. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_render.py +0 -238
  77. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_render_encoding.py +0 -70
  78. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_rollup.py +0 -263
  79. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_snapshot.py +0 -63
  80. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_sync.py +0 -52
  81. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_transcript_encoding.py +0 -109
  82. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_transcripts.py +0 -151
  83. package/integrations/openclaw/skills/agent-cost-report/scripts/tests/test_wins.py +0 -222
  84. package/plugin/skills/agent-cost-report/CHECKSUMS.txt +0 -50
  85. package/plugin/skills/agent-cost-report/SKILL.md +0 -241
  86. package/plugin/skills/agent-cost-report/scripts/acr/__init__.py +0 -1
  87. package/plugin/skills/agent-cost-report/scripts/acr/behavior.py +0 -303
  88. package/plugin/skills/agent-cost-report/scripts/acr/classify.py +0 -83
  89. package/plugin/skills/agent-cost-report/scripts/acr/costs.py +0 -119
  90. package/plugin/skills/agent-cost-report/scripts/acr/devices.py +0 -88
  91. package/plugin/skills/agent-cost-report/scripts/acr/evidence.py +0 -160
  92. package/plugin/skills/agent-cost-report/scripts/acr/labels.py +0 -115
  93. package/plugin/skills/agent-cost-report/scripts/acr/measure.py +0 -89
  94. package/plugin/skills/agent-cost-report/scripts/acr/mistakes.py +0 -310
  95. package/plugin/skills/agent-cost-report/scripts/acr/patterns.py +0 -391
  96. package/plugin/skills/agent-cost-report/scripts/acr/period.py +0 -98
  97. package/plugin/skills/agent-cost-report/scripts/acr/prices.py +0 -95
  98. package/plugin/skills/agent-cost-report/scripts/acr/render.py +0 -586
  99. package/plugin/skills/agent-cost-report/scripts/acr/rollup.py +0 -355
  100. package/plugin/skills/agent-cost-report/scripts/acr/rules.json +0 -4
  101. package/plugin/skills/agent-cost-report/scripts/acr/rules.py +0 -76
  102. package/plugin/skills/agent-cost-report/scripts/acr/sample.py +0 -45
  103. package/plugin/skills/agent-cost-report/scripts/acr/snapshot.py +0 -53
  104. package/plugin/skills/agent-cost-report/scripts/acr/sync.py +0 -79
  105. package/plugin/skills/agent-cost-report/scripts/acr/transcripts.py +0 -126
  106. package/plugin/skills/agent-cost-report/scripts/acr/wins.py +0 -341
  107. package/plugin/skills/agent-cost-report/scripts/acr.py +0 -219
  108. package/plugin/skills/agent-cost-report/scripts/tests/_db.py +0 -49
  109. package/plugin/skills/agent-cost-report/scripts/tests/_paths.py +0 -9
  110. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/codex/sessions/2026/09/18/rollout-codex-1.jsonl +0 -4
  111. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/make_report_fixtures.py +0 -46
  112. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/prices_sample.json +0 -6
  113. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-a/sess-aaaa.jsonl +0 -8
  114. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/projects/-workspace-proj-b/sess-bbbb.jsonl +0 -1
  115. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-1day.json +0 -823
  116. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-30day.json +0 -2213
  117. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-7day.json +0 -1252
  118. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-empty.json +0 -536
  119. package/plugin/skills/agent-cost-report/scripts/tests/fixtures/report-session.json +0 -662
  120. package/plugin/skills/agent-cost-report/scripts/tests/test_behavior.py +0 -299
  121. package/plugin/skills/agent-cost-report/scripts/tests/test_devices.py +0 -99
  122. package/plugin/skills/agent-cost-report/scripts/tests/test_labels.py +0 -96
  123. package/plugin/skills/agent-cost-report/scripts/tests/test_measure.py +0 -120
  124. package/plugin/skills/agent-cost-report/scripts/tests/test_pdf_windows.py +0 -54
  125. package/plugin/skills/agent-cost-report/scripts/tests/test_period.py +0 -137
  126. package/plugin/skills/agent-cost-report/scripts/tests/test_prices.py +0 -126
  127. package/plugin/skills/agent-cost-report/scripts/tests/test_render.py +0 -238
  128. package/plugin/skills/agent-cost-report/scripts/tests/test_render_encoding.py +0 -70
  129. package/plugin/skills/agent-cost-report/scripts/tests/test_rollup.py +0 -263
  130. package/plugin/skills/agent-cost-report/scripts/tests/test_snapshot.py +0 -63
  131. package/plugin/skills/agent-cost-report/scripts/tests/test_sync.py +0 -52
  132. package/plugin/skills/agent-cost-report/scripts/tests/test_transcript_encoding.py +0 -109
  133. package/plugin/skills/agent-cost-report/scripts/tests/test_transcripts.py +0 -151
  134. package/plugin/skills/agent-cost-report/scripts/tests/test_wins.py +0 -222
  135. package/plugin/skills/babysit/SKILL.md +0 -87
  136. package/plugin/skills/design-is/SKILL.md +0 -312
  137. package/plugin/skills/do/SKILL.md +0 -45
  138. package/plugin/skills/learn-codebase/SKILL.md +0 -21
  139. package/plugin/skills/make-plan/SKILL.md +0 -67
  140. package/plugin/skills/oh-my-issues/SKILL.md +0 -226
  141. package/plugin/skills/pathfinder/SKILL.md +0 -111
  142. package/plugin/skills/screenpipe/SKILL.md +0 -78
  143. package/plugin/skills/screenpipe/screenpipe.mjs +0 -196
  144. package/plugin/skills/smart-explore/SKILL.md +0 -193
  145. package/plugin/skills/standup/SKILL.md +0 -142
  146. package/plugin/skills/standup/agent-brief.md +0 -47
  147. package/plugin/skills/standup/standup.mjs +0 -586
  148. package/plugin/skills/what-the/SKILL.md +0 -6
  149. package/plugin/skills/wowerpoint/SKILL.md +0 -205
@@ -1,126 +0,0 @@
1
- """Measured token usage from Claude Code transcripts (plan Phase 1.4).
2
-
3
- Claude Code reader copied verbatim from parse_transcripts.py:7-38
4
- and the Codex reader from :39-57, then extended (plan 1.4) with `cwd`, `cache_write_5m`
5
- (read directly from cache_creation.ephemeral_5m_input_tokens when present, else
6
- cache_write - cache_write_1h as weekly_report.py:94) and `device = "local"`.
7
-
8
- Every transcript is streamed line by line; no file is ever read whole.
9
- """
10
- import datetime as dt
11
- import glob
12
- import json
13
- import os
14
- import platform
15
-
16
- from .period import PT, session_block
17
-
18
- CLAUDE_GLOB = os.path.join("~", ".claude", "projects", "**", "*.jsonl")
19
- CODEX_GLOB = os.path.join("~", ".codex", "sessions", "**", "*.jsonl")
20
- DEVICE = "local"
21
-
22
-
23
- def host_label():
24
- """platform.node(); never a device UUID from settings."""
25
- return platform.node()
26
-
27
-
28
- def ts(x):
29
- return dt.datetime.fromisoformat(x.replace("Z", "+00:00"))
30
-
31
-
32
- def _project_dir(f, pattern):
33
- """Project dir = path segment after `/.claude/projects/` (parse_transcripts.py:20).
34
- For a non-default glob root (tests) the first segment relative to the root is the same thing."""
35
- if "/.claude/projects/" in f:
36
- return f.split("/.claude/projects/")[1].split("/")[0]
37
- root = os.path.expanduser(pattern).split("**")[0]
38
- return os.path.relpath(f, root).split(os.sep)[0]
39
-
40
-
41
- def collect_claude(s, e, session=None, pattern=CLAUDE_GLOB):
42
- """Claude Code assistant lines in [s, e) deduped on (message.id, requestId).
43
-
44
- s/e: aware datetimes (end exclusive), or None when `session` is given (no period filter,
45
- plan 1.2). Returns (rows, stats)."""
46
- seen = set(); rows = []; dropped = 0
47
- files = [f for f in glob.glob(os.path.expanduser(pattern), recursive=True)
48
- if s is None or dt.datetime.fromtimestamp(os.path.getmtime(f), PT) >= s]
49
- for f in files:
50
- proj = _project_dir(f, pattern)
51
- with open(f, encoding="utf-8", errors="ignore") as fh:
52
- for line in fh:
53
- if '"usage"' not in line: continue
54
- if session is not None and session not in line: continue
55
- try: d = json.loads(line)
56
- except Exception: continue
57
- m = d.get("message") or {}
58
- u = m.get("usage")
59
- if d.get("type") != "assistant" or not u or not d.get("timestamp"): continue
60
- sid = d.get("sessionId") or d.get("session_id")
61
- if session is not None and sid != session: continue
62
- t = ts(d["timestamp"])
63
- if s is not None and not (s <= t < e): continue
64
- key = (m.get("id"), d.get("requestId"))
65
- if key in seen: dropped += 1; continue
66
- seen.add(key)
67
- cc = u.get("cache_creation") or {}
68
- cache_write = u.get("cache_creation_input_tokens", 0) or 0
69
- cache_write_1h = cc.get("ephemeral_1h_input_tokens", 0) or 0
70
- cw5 = cc.get("ephemeral_5m_input_tokens")
71
- cache_write_5m = cw5 if cw5 is not None else cache_write - cache_write_1h
72
- rows.append(dict(src="claude-code", file=f, dir=proj, cwd=d.get("cwd"), session=sid,
73
- sidechain=bool(d.get("isSidechain")), model=m.get("model"), ts=t.astimezone(PT).isoformat(),
74
- input=u.get("input_tokens", 0) or 0, output=u.get("output_tokens", 0) or 0,
75
- cache_write=cache_write, cache_write_1h=cache_write_1h, cache_write_5m=cache_write_5m,
76
- cache_read=u.get("cache_read_input_tokens", 0) or 0,
77
- cost_usd_reported=d.get("costUSD"), device=DEVICE))
78
- return rows, dict(files_seen=len(files), dedup_dropped=dropped)
79
-
80
-
81
- def collect_codex(s, e, session=None, pattern=CODEX_GLOB):
82
- """Codex: token_count events carry a cumulative total_token_usage per session file (parse_transcripts.py:39-57
83
- took the last one). Each event becomes one row holding the growth since the previous event, so usage before
84
- the window is subtracted and every day is charged its own tokens; a session that starts before the window
85
- contributes only its in-window growth."""
86
- rows = []; files_seen = 0
87
- for f in glob.glob(os.path.expanduser(pattern), recursive=True):
88
- if s is not None and dt.datetime.fromtimestamp(os.path.getmtime(f), PT) < s: continue
89
- if session is not None and os.path.basename(f) != session: continue
90
- files_seen += 1
91
- prev = dict(input_tokens=0, cached_input_tokens=0, output_tokens=0); model = None
92
- with open(f, encoding="utf-8", errors="ignore") as fh:
93
- for line in fh:
94
- try: d = json.loads(line)
95
- except Exception: continue
96
- p = d.get("payload") or {}
97
- if d.get("type") == "turn_context": model = p.get("model", model)
98
- if p.get("type") == "token_count" and p.get("info"):
99
- t = ts(d["timestamp"]); u = p["info"].get("total_token_usage") or {}
100
- cur = {k: int(u.get(k, 0) or 0) for k in prev}
101
- delta = {k: max(cur[k] - prev[k], 0) for k in prev}; prev = cur
102
- if not any(delta.values()) or (s is not None and not (s <= t < e)): continue
103
- rows.append(dict(src="codex", file=f, dir="codex", cwd=None, session=os.path.basename(f), sidechain=False, model=model,
104
- ts=t.astimezone(PT).isoformat(), input=max(delta["input_tokens"] - delta["cached_input_tokens"], 0),
105
- output=delta["output_tokens"], cache_write=0, cache_write_1h=0, cache_write_5m=0, cache_read=delta["cached_input_tokens"],
106
- cost_usd_reported=None, device=DEVICE))
107
- return rows, dict(files_seen=files_seen)
108
-
109
-
110
- def collect(window=None, session=None, claude_glob=CLAUDE_GLOB, codex_glob=CODEX_GLOB, window_block=None):
111
- """Run both readers. `window` is a period.Window, or None with `session` set.
112
- Returns the usage.json document: {window, collector, rows}."""
113
- if window is None and session is None:
114
- raise ValueError("collect needs a window or a session id")
115
- s, e = (window.start, window.end) if window is not None else (None, None)
116
- claude_rows, cstats = collect_claude(s, e, session=session, pattern=claude_glob)
117
- codex_rows, xstats = collect_codex(s, e, session=session, pattern=codex_glob)
118
- rows = claude_rows + codex_rows
119
- if window_block is None:
120
- window_block = window.block() if window is not None else session_block(session)
121
- wblock = window_block
122
- collector = dict(host_label=host_label(), glob=[claude_glob, codex_glob],
123
- files_seen=cstats["files_seen"] + xstats["files_seen"],
124
- files_seen_by_src={"claude-code": cstats["files_seen"], "codex": xstats["files_seen"]},
125
- rows=len(rows), dedup_dropped=cstats["dedup_dropped"], window=wblock)
126
- return dict(window=wblock, collector=collector, rows=rows)
@@ -1,341 +0,0 @@
1
- """Wins and what each cost (plan Phase 2.8, G11/G12/G13).
2
-
3
- A win is a merged PR or a published version (praise wins are a Phase 2B hook, see detect_praise).
4
- Sources: successful `gh pr merge` / `npm publish` / `gh release create` / tag-push tool calls found by
5
- streaming the window's transcripts, and ship observations (evidence.SHIP_RE, weekly_report.py:65,74-75)
6
- that name a PR number or a published version. Deduped by key repo#number or package@version.
7
- Optional read-only confirmation: `gh pr view <n> --repo <repo> --json mergedAt,url,title,commits`
8
- (one call per candidate PR; never a write command). mergedAt must fall in the window.
9
-
10
- Cost: "unmeasured" (every usd null) unless the PR's commits carry `Claude-Session: <bare id>`
11
- trailers (R4, G13: a `Session:` line in a PR body is ignored; non-bare values rejected). Then each turn
12
- of a linked session is assigned to the first linked win merged after it, so no turn counts twice.
13
- There is no lineage fallback (G11). Titles come from the PR, the package, or the observation.
14
- """
15
- import collections
16
- import datetime as dt
17
- import json
18
- import re
19
- import subprocess
20
-
21
- from .labels import active_minutes
22
- from .period import PT
23
-
24
- KEYS = ("gh pr merge", "npm publish", "gh release create", "git push")
25
- SEGMENT_SPLIT = re.compile(r"[;&|\n]")
26
- MERGE_CMD = re.compile(r"^gh\s+pr\s+merge\b(?P<rest>.*)")
27
- PUBLISH_CMD = re.compile(r"^npm\s+publish\b")
28
- RELEASE_CMD = re.compile(r"^gh\s+release\s+create\s+(?P<tag>[\w.-]+)")
29
- TAG_PUSH_CMD = re.compile(r"^git\s+push\b(?P<rest>.*)")
30
- FAILED = re.compile(r"^\W{0,3}(error|fatal|failed|graphql:|npm err!|could not|permission denied|rejected|x pull request)", re.I | re.M)
31
- PR_NUM = re.compile(r"/pull/(\d+)|(?<![\w.#-])(\d+)(?![\w.-])")
32
- REPO_FLAG = re.compile(r"(?:--repo|-R)[ =](\S+)")
33
- PR_IN_TITLE = re.compile(r"(?:\bPR|pull request)\s*#\s*(\d+)", re.I)
34
- PUBLISH_TITLE = re.compile(r"\b(published|released|npm publish|tagged)\b", re.I)
35
- PKG_VER = re.compile(r"(@?[\w][\w./-]*)@(\d+\.\d+\.\d+[\w.-]*)")
36
- VERSION = re.compile(r"\bv?(\d+\.\d+\.\d+(?:[-.][\w.]+)?)\b")
37
- REPO_URL = re.compile(r"github\.com[:/]([\w.-]+)/([\w.-]+?)(?:\.git)?(?:[/\s]|$)")
38
- TRAILER = re.compile(r"^Claude-Session:[ \t]*(.*?)[ \t]*$", re.M)
39
- BARE_ID = re.compile(r"^[A-Za-z0-9][A-Za-z0-9._-]*$")
40
- LINKED_LABEL = "ESTIMATED · session-linked"
41
- ATTRIBUTION = "Claude-Session commit trailers (R4); each turn assigned to the first linked win merged after it"
42
-
43
-
44
- def classify_command(cmd):
45
- """The first merge/publish verb that starts a command segment (heredoc bodies and quoted prose never count)."""
46
- for seg in SEGMENT_SPLIT.split(cmd.split("<<")[0]):
47
- seg = seg.strip().lstrip("( ").strip()
48
- m = MERGE_CMD.match(seg)
49
- if m:
50
- rest = m.group("rest"); n = next((a or b for a, b in PR_NUM.findall(rest) if a or b), None)
51
- rf = REPO_FLAG.search(rest); ru = REPO_URL.search(rest)
52
- repo = rf.group(1) if rf else (f"{ru.group(1)}/{ru.group(2)}" if ru else None)
53
- return dict(kind="pr", number=int(n) if n else None, repo=repo, cmd=cmd[:200])
54
- if PUBLISH_CMD.match(seg): return dict(kind="publish", how="npm publish", cmd=cmd[:200])
55
- m = RELEASE_CMD.match(seg)
56
- if m: return dict(kind="publish", how="gh release create", tag=m.group("tag"), cmd=cmd[:200])
57
- m = TAG_PUSH_CMD.match(seg)
58
- if m:
59
- rest = m.group("rest"); v = VERSION.search(rest)
60
- if "--tags" in rest or v: return dict(kind="publish", how="git push tag", tag=v.group(0) if v else None, cmd=cmd[:200])
61
- return None
62
-
63
-
64
- def _result_text(c, tur):
65
- parts = []
66
- if isinstance(tur, dict): parts += [str(tur.get("stdout") or ""), str(tur.get("stderr") or "")]
67
- cc = c.get("content")
68
- if isinstance(cc, str): parts.append(cc)
69
- elif isinstance(cc, list): parts += [str(x.get("text") or "") for x in cc if isinstance(x, dict)]
70
- return "\n".join(parts)
71
-
72
-
73
- def _ts_ms(iso):
74
- try: return int(dt.datetime.fromisoformat(iso.replace("Z", "+00:00")).timestamp() * 1000)
75
- except Exception: return None
76
-
77
-
78
- def scan_transcripts(files, s_ms=None, e_ms=None):
79
- """Stream each transcript line by line; yield successful merge/publish tool calls inside the window."""
80
- for f in files:
81
- pending = {}
82
- try: fh = open(f, encoding="utf-8", errors="ignore")
83
- except OSError: continue
84
- with fh:
85
- for line in fh:
86
- if not pending and not any(k in line for k in KEYS): continue
87
- if '"tool_use"' not in line and '"tool_result"' not in line: continue
88
- try: d = json.loads(line)
89
- except Exception: continue
90
- content = (d.get("message") or {}).get("content")
91
- if not isinstance(content, list): continue
92
- if d.get("type") == "assistant":
93
- for c in content:
94
- if isinstance(c, dict) and c.get("type") == "tool_use" and c.get("name") == "Bash":
95
- cand = classify_command(str((c.get("input") or {}).get("command") or ""))
96
- if cand:
97
- cand.update(tool_use_id=c.get("id"), ts_ms=_ts_ms(d.get("timestamp") or ""), session=d.get("sessionId"),
98
- cwd=d.get("cwd"), file=f, source="transcript")
99
- pending[c.get("id")] = cand
100
- elif d.get("type") == "user" and pending:
101
- for c in content:
102
- if isinstance(c, dict) and c.get("type") == "tool_result" and c.get("tool_use_id") in pending:
103
- cand = pending.pop(c["tool_use_id"]); text = _result_text(c, d.get("toolUseResult"))
104
- if c.get("is_error") or FAILED.search(text): continue
105
- if cand["ts_ms"] is None or (s_ms is not None and not (s_ms <= cand["ts_ms"] < e_ms)): continue
106
- cand["result"] = text[:2000]; yield cand
107
-
108
-
109
- def parse_repo(url):
110
- m = REPO_URL.search(url or "")
111
- return f"{m.group(1)}/{m.group(2)}" if m else None
112
-
113
-
114
- def git_remote(cwd):
115
- """Read-only: `git -C cwd remote get-url origin`; None when cwd is gone or not a repo."""
116
- try:
117
- r = subprocess.run(["git", "-C", cwd, "remote", "get-url", "origin"], capture_output=True, text=True, encoding="utf-8", errors="replace", timeout=5)
118
- return parse_repo(r.stdout) if r.returncode == 0 else None
119
- except Exception: return None
120
-
121
-
122
- def gh_pr_view(number, repo):
123
- """Read-only: gh pr view <n> --repo <repo> --json mergedAt,url,title,commits. Returns dict or None."""
124
- try:
125
- r = subprocess.run(["gh", "pr", "view", str(number), "--repo", repo, "--json", "mergedAt,url,title,commits"],
126
- capture_output=True, text=True, encoding="utf-8", errors="replace", timeout=25)
127
- return json.loads(r.stdout) if r.returncode == 0 and r.stdout.strip() else None
128
- except Exception: return None
129
-
130
-
131
- WINS_REPOS = ("imogenlabs/memindex",)
132
- GH_SOURCE = "github_merged_prs"
133
- GH_LIST_LIMIT = 200
134
-
135
-
136
- def _gh_status(returncode, stderr):
137
- """Map a failed `gh` call to an honest status; never a number."""
138
- t = (stderr or "").lower()
139
- if "rate limit" in t or "api rate limit" in t or "secondary rate" in t: return "rate-limited"
140
- if "auth login" in t or "authentication" in t or "not logged in" in t or "http 401" in t or "bad credentials" in t: return "unauthenticated"
141
- return f"gh error (exit {returncode})"
142
-
143
-
144
- def gh_merged_prs(repo, s_ms, e_ms, run=subprocess.run):
145
- """Read-only: gh pr list --repo <repo> --state merged --json number,title,mergedAt,url, searched by merge date
146
- (UTC days covering the PT window) and then filtered by mergedAt inside [s_ms, e_ms). Returns (rows, status)
147
- with status "ok" or a reason: "gh missing" | "unauthenticated" | "rate-limited" | "gh error (exit n)"."""
148
- a = dt.datetime.fromtimestamp(s_ms / 1000, dt.timezone.utc).date() - dt.timedelta(days=1)
149
- b = dt.datetime.fromtimestamp(e_ms / 1000, dt.timezone.utc).date() + dt.timedelta(days=1)
150
- cmd = ["gh", "pr", "list", "--repo", repo, "--state", "merged", "--limit", str(GH_LIST_LIMIT), "--json", "number,title,mergedAt,url",
151
- "--search", f"merged:{a.isoformat()}..{b.isoformat()}"]
152
- try:
153
- r = run(cmd, capture_output=True, text=True, encoding="utf-8", errors="replace", timeout=40)
154
- except FileNotFoundError: return [], "gh missing"
155
- except (subprocess.TimeoutExpired, OSError) as ex: return [], f"gh error ({type(ex).__name__})"
156
- if r.returncode != 0: return [], _gh_status(r.returncode, r.stderr)
157
- try: rows = json.loads(r.stdout or "[]")
158
- except ValueError: return [], "gh error (unparseable output)"
159
- out = []
160
- for x in rows:
161
- if not x.get("mergedAt") or not x.get("number"): continue
162
- ms = _ts_ms(x["mergedAt"])
163
- if s_ms <= ms < e_ms: out.append(dict(number=int(x["number"]), title=x.get("title") or "", url=x.get("url"), ts_ms=ms))
164
- return out, "ok"
165
-
166
-
167
- def gh_list_candidates(repos, s_ms, e_ms, lister=gh_merged_prs):
168
- """Candidates from the merged-PR list per repo plus a per-repo status for the report."""
169
- cands = []; status = {}
170
- for repo in repos:
171
- rows, st = lister(repo, s_ms, e_ms)
172
- status[repo] = dict(status=st, found=len(rows) if st == "ok" else None)
173
- for x in rows:
174
- cands.append(dict(kind="pr", source=GH_SOURCE, repo=repo, number=x["number"], title=x["title"], url=x["url"], ts_ms=x["ts_ms"],
175
- project=repo.split("/")[-1], sessions=[], evidence_ids=[], gh_listed=True))
176
- return cands, status
177
-
178
-
179
- def session_trailers(message):
180
- """Bare content_session_ids named by `Claude-Session:` trailers in one commit message (G13).
181
- Non-bare values (URLs, ids with spaces or parentheses, anything else) are rejected; `Session:`
182
- lines and PR bodies are never read."""
183
- out = []
184
- for v in TRAILER.findall(message or ""):
185
- if v and BARE_ID.match(v) and v not in out: out.append(v)
186
- return out
187
-
188
-
189
- def detect_praise(*_args, **_kw):
190
- """Phase 2B hook (plan 2B.1 human-vs-bot tagger, 2B.4 sarcasm filter). Until those exist no user
191
- prompt is read for praise, so this returns [] and wins.praise_n is 0. Praise wins are never costed."""
192
- return []
193
-
194
-
195
- def _pt_iso(ms):
196
- return dt.datetime.fromtimestamp(ms / 1000, PT).strftime("%Y-%m-%dT%H:%M%z")
197
-
198
-
199
- def observation_candidates(ship_obs, project_repo):
200
- """ship_obs: dicts(id, title, ts_ms, project, memory_session_id, content_session_id)."""
201
- for o in ship_obs:
202
- t = o["title"] or ""; base = (o["project"] or "").split("/")[0]
203
- m = PR_IN_TITLE.search(t)
204
- if m:
205
- yield dict(kind="pr", number=int(m.group(1)), repo=project_repo(o["project"]), project=o["project"], ts_ms=o["ts_ms"],
206
- title=t, evidence_ids=[o["id"]], sessions=[o["content_session_id"]], source="ship_observation"); continue
207
- if PUBLISH_TITLE.search(t):
208
- pv = PKG_VER.search(t); v = VERSION.search(t)
209
- if pv: key = f"{pv.group(1)}@{pv.group(2)}"
210
- elif v and base: key = f"{base}@{v.group(1)}"
211
- else: continue
212
- yield dict(kind="publish", key=key, project=o["project"], ts_ms=o["ts_ms"], title=t, evidence_ids=[o["id"]],
213
- sessions=[o["content_session_id"]], source="ship_observation")
214
-
215
-
216
- def _key(c, dropped):
217
- if c["kind"] == "pr":
218
- if not c.get("number"): dropped.append(dict(source=c["source"], reason="gh pr merge without a PR number", cmd=c.get("cmd"))); return None
219
- return f"{c.get('repo') or (c.get('project') or '?').split('/')[0]}#{c['number']}"
220
- if c.get("key"): return c["key"]
221
- text = c.get("result") or ""; pv = PKG_VER.search(text)
222
- if c.get("how") == "npm publish":
223
- if pv: return f"{pv.group(1)}@{pv.group(2)}"
224
- dropped.append(dict(source=c["source"], reason="npm publish result names no package@version", cmd=c.get("cmd"))); return None
225
- if c.get("tag"): return f"{c.get('repo') or c.get('project') or '?'}@{c['tag']}"
226
- dropped.append(dict(source=c["source"], reason="tag push without an identifiable tag", cmd=c.get("cmd"))); return None
227
-
228
-
229
- def build(files, s_ms, e_ms, ship_obs, sessions_by_cs, rows_by_cs, ratio, days, gh=gh_pr_view, remote=git_remote, use_gh=True,
230
- repos=WINS_REPOS, gh_list=gh_merged_prs):
231
- """Returns (wins_block, wins_by_day). sessions_by_cs: cs -> dict(project, memory_session_id, has_transcript,
232
- observer_tokens, cwd); rows_by_cs: cs -> usage rows with ts_ms and x1e6 (None when unpriced)."""
233
- repo_cache = {}
234
- def project_repo(project):
235
- """Repo of a project from `git remote` of any transcript cwd in it; a worktree (`repo/wt`) falls back to its base project."""
236
- if project in repo_cache: return repo_cache[project]
237
- repo = None; base = (project or "").split("/")[0]
238
- for cs, s in sessions_by_cs.items():
239
- if (s.get("project") or "").split("/")[0] == base and s.get("cwd"):
240
- repo = remote(s["cwd"]) if remote else None
241
- if repo: break
242
- repo_cache[project] = repo; return repo
243
- cands = []; dropped = []; gh_unconfirmed = []
244
- for c in scan_transcripts(files, s_ms, e_ms):
245
- s = sessions_by_cs.get(c["session"]) or {}
246
- c["project"] = s.get("project") or c.get("cwd"); c["sessions"] = [c["session"]]; c["evidence_ids"] = []
247
- if c["kind"] == "pr" and not c.get("repo"): c["repo"] = (remote(c["cwd"]) if remote and c.get("cwd") else None) or project_repo(c["project"])
248
- cands.append(c)
249
- cands += list(observation_candidates(ship_obs, project_repo))
250
- gh_status = {}
251
- if use_gh and gh_list and s_ms is not None and e_ms is not None:
252
- gh_cands, gh_status = gh_list_candidates(repos, s_ms, e_ms, lister=gh_list); cands += gh_cands
253
- elif repos: gh_status = {r: dict(status="unavailable (gh disabled for this run)", found=None) for r in repos}
254
- merged = collections.OrderedDict()
255
- for c in cands:
256
- key = _key(c, dropped)
257
- if key is None: continue
258
- w = merged.get(key)
259
- if w is None:
260
- w = merged[key] = dict(win_id=None, kind=c["kind"], key=key, title=c.get("title") or key, url=None, ts_ms=c["ts_ms"],
261
- project=c.get("project"), repo=c.get("repo"), number=c.get("number"), evidence_ids=[], sessions=[],
262
- sources=[], confirmed=False, linked_sessions=[], gh_listed=False)
263
- for eid in c.get("evidence_ids", []):
264
- if eid not in w["evidence_ids"]: w["evidence_ids"].append(eid)
265
- for sid in c.get("sessions", []):
266
- if sid and sid not in w["sessions"]: w["sessions"].append(sid)
267
- if c["source"] not in w["sources"]: w["sources"].append(c["source"])
268
- if c.get("gh_listed"):
269
- w["gh_listed"] = True; w["confirmed"] = True; w["ts_ms"] = c["ts_ms"]; w["url"] = w["url"] or c.get("url"); w["title"] = c.get("title") or w["title"]
270
- elif c["ts_ms"] and (w["ts_ms"] is None or c["ts_ms"] < w["ts_ms"]): w["ts_ms"] = c["ts_ms"]
271
- items = []
272
- for key, w in merged.items():
273
- if w["kind"] == "pr" and use_gh and gh and w.get("repo") and w.get("number") and not (w["gh_listed"] and w["sources"] == [GH_SOURCE]):
274
- info = gh(w["number"], w["repo"])
275
- if info is not None:
276
- if not info.get("mergedAt"): dropped.append(dict(key=key, reason="gh pr view: not merged")); continue
277
- w["ts_ms"] = _ts_ms(info["mergedAt"]); w["url"] = info.get("url"); w["title"] = info.get("title") or w["title"]
278
- w["confirmed"] = True; w["sources"].append("gh_confirm")
279
- for cm in info.get("commits") or []:
280
- for sid in session_trailers("\n".join([cm.get("messageHeadline") or "", cm.get("messageBody") or ""])):
281
- if sid not in w["linked_sessions"]: w["linked_sessions"].append(sid)
282
- else:
283
- gh_unconfirmed.append(dict(key=key, reason="gh pr view returned nothing (gh error, no access, or PR not found); kept unconfirmed"))
284
- if w["ts_ms"] is None: dropped.append(dict(key=key, reason="no timestamp")); continue
285
- if s_ms is not None and not (s_ms <= w["ts_ms"] < e_ms): dropped.append(dict(key=key, reason="outside window", ts_pt=_pt_iso(w["ts_ms"]))); continue
286
- items.append(w)
287
- items.sort(key=lambda w: w["ts_ms"])
288
- for i, w in enumerate(items, 1): w["win_id"] = f"W-{i}"
289
- attribute(items, sessions_by_cs, rows_by_cs, ratio)
290
- praise = detect_praise()
291
- out = []
292
- for w in items:
293
- out.append(dict(win_id=w["win_id"], kind=w["kind"], key=w["key"], title=w["title"], url=w["url"], ts_pt=_pt_iso(w["ts_ms"]),
294
- day_pt=dt.datetime.fromtimestamp(w["ts_ms"] / 1000, PT).date().isoformat(), project=w["project"],
295
- evidence_ids=w["evidence_ids"], sessions=w["sessions"], sources=w["sources"], confirmed=w["confirmed"],
296
- linked_sessions=w["linked_sessions"], usd=w["usd"], cost_status=w["cost_status"], cost_basis=w["cost_basis"],
297
- cost_label=w["cost_label"], tokens=w["tokens"], sessions_n=w["sessions_n"], turns_n=w["turns_n"],
298
- active_minutes=w["active_minutes"]))
299
- linked = any(w["cost_status"] == "session_linked" for w in items)
300
- total = round(sum(w["usd"] for w in items if w["usd"] is not None), 2) if linked else None
301
- block = dict(items=out, cost_status="session_linked" if linked else "unmeasured", attribution_method=ATTRIBUTION if linked else None,
302
- unattributed_usd=None, total_attributed_usd=total, praise_n=len(praise), dropped=dropped, gh_unconfirmed=gh_unconfirmed,
303
- sources=dict(transcripts="box transcripts (gh pr merge / npm publish / release / tag push)", ship_observations="memindex ship observations",
304
- **{GH_SOURCE: dict(label="GitHub merged PRs (read-only gh pr list)", repos={r: v for r, v in gh_status.items()},
305
- status=("ok" if gh_status and all(v["status"] == "ok" for v in gh_status.values()) else "unavailable" if gh_status else "off"),
306
- found=sum(v["found"] or 0 for v in gh_status.values()) if gh_status else 0)}),
307
- note="finished line items (shipped/completed) are outcomes, not wins (G12); praise detection is a Phase 2B hook")
308
- by_day = [dict(day_pt=d, count=sum(1 for w in out if w["day_pt"] == d),
309
- usd=(round(sum(w["usd"] for w in out if w["day_pt"] == d and w["usd"] is not None), 2) if linked else None),
310
- win_ids=[w["win_id"] for w in out if w["day_pt"] == d]) for d in days]
311
- return block, by_day
312
-
313
-
314
- def attribute(items, sessions_by_cs, rows_by_cs, ratio):
315
- """Assign each turn of a linked session to the first linked win merged after it (no double counting).
316
- Replica sessions (no transcript on this box) add an extrapolated portion only when the win is linked
317
- and a ratio exists; the item is then labeled cost_basis extrapolated. Unlinked wins stay unmeasured."""
318
- by_session = collections.defaultdict(list)
319
- for w in items:
320
- w.update(usd=None, cost_status="unmeasured", cost_basis=None, cost_label="unmeasured (session not linked to PR)",
321
- tokens=None, sessions_n=0, turns_n=0, active_minutes=None, _x1e6=0, _tok=collections.Counter(), _stamps=[])
322
- for sid in w["linked_sessions"]: by_session[sid].append(w)
323
- for sid, ws in by_session.items():
324
- ws.sort(key=lambda w: w["ts_ms"])
325
- for w in ws: w["sessions_n"] += 1
326
- for r in rows_by_cs.get(sid, []):
327
- tgt = next((w for w in ws if w["ts_ms"] >= r["ts_ms"]), None)
328
- if tgt is None: continue
329
- tgt["turns_n"] += 1; tgt["_stamps"].append(r["ts_ms"])
330
- if r.get("x1e6") is not None: tgt["_x1e6"] += r["x1e6"]
331
- for f in ("input", "output", "cache_write_5m", "cache_write_1h", "cache_read"): tgt["_tok"][f] += r[f]
332
- s = sessions_by_cs.get(sid) or {}
333
- if not s.get("has_transcript") and ratio is not None and s.get("observer_tokens"):
334
- ws[-1]["_x1e6"] += int(round(s["observer_tokens"] * ratio)); ws[-1]["cost_basis"] = "extrapolated"
335
- for w in items:
336
- if w["linked_sessions"]:
337
- w["usd"] = round(w["_x1e6"] / 1e6, 2); w["cost_status"] = "session_linked"
338
- w["cost_basis"] = w["cost_basis"] or "estimated_usage"
339
- w["cost_label"] = LINKED_LABEL + (" · EXTRAPOLATED (low confidence)" if w["cost_basis"] == "extrapolated" else "")
340
- w["tokens"] = dict(w["_tok"]); w["active_minutes"] = round(active_minutes(w["_stamps"]), 1)
341
- for k in ("_x1e6", "_tok", "_stamps"): w.pop(k, None)