tmpl/multi-agent-mux
1
0 Fork 0
Code Issues 9 Pull Requests Actions Packages Projects Releases Wiki Activity

fix(agents): pin post-discovered session ids on a verified completed; agy never takes another workspace's or session's conversation (Issue #40)

Browse Source 
- BaseAgentAdapter: id_lifecycle (pre_assigned: claude, grok / post_discovered)
  and capture_active_session_id - exactly one verified candidate newer than the
  session that no other row records, else a reason (no-candidate, ambiguous, ...).
- agy: verify_artifact (discover) rejects an id last_conversations.json maps to
  another workspace and always excludes ids other rows record (a cache hit no
  longer bypasses it). capture uses the cache only as a proposal: the
  conversation's presence lock or DB must be held by the row's pane process or
  a descendant (lsof/ps); anything unprovable is not pinned.
- lib_py/agents/autopin.py: handle_completed, run by the side that verified the
  job's completed - the delegate-job wrapper (direct), create --onboard and the
  loop's wait_for_job (recovery, as a subprocess: G9). Local herdr jobs only;
  one (job, dispatch token, attempt) is handled once (job-scoped flock + marker
  in the job dir); writes go through atomic_dump_yaml; a pin records
  session_id_source=discovered / session_id_verified=true; a failure leaves the
  reason on the row and never changes the job's result.
- reconcile.sh drift C for agy, hermes, opencode and codex calls the same
  capture (the agy *.db glob is gone); _pin_and_verify_resume uses pin_row.
- Tests: tests/test_issue40_auto_pin.py; issue64 :407 now expects False; the
  static reconcile-source checks follow the move to the adapters; the two tier2
  drift-C component tests create with --no-onboard (onboarding now pins).

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
This commit is contained in:
GodopuandClaude Opus 5.5 committed 2026-10-20 23:40:03 +09:00
1 parent ff8b381520
commit feccfd5e03
19 files changed
+1128 -338
Download Patch File Download Diff File Expand all files Collapse all files

No files matched your search

.agents/skills/multi-agent-mux-monitor/scripts/reconcile.sh
+23 -305
View File
@@ -420,7 +420,7 @@ import yaml
from lib_py.verify_session import verify_session_uuid, workspace_key
from lib_py.agents.registry import get_adapter
from lib_py.agents.base import DiscoveryContext, _ro_connect
from lib_py.agents import autopin
def _slug(path):
if not path:
@@ -554,17 +554,9 @@ def pane_meta(session, srv):
from lib_py.agents.registry import get_adapter, own_key as _get_own_key, agent_of_row
def _pin_and_verify_resume(s, agent, cwd, uuid, degraded=False):
own_key_field = _get_own_key(agent)
if own_key_field:
s[own_key_field] = uuid
s['last_visible_status'] = 'pinned'
resume_cmd = ['bash', os.path.join(skills_dir, 'multi-agent-mux-resume', 'scripts', 'resume_session.sh'),
'--workspace', cwd, '--agent', agent, '--session', s['name'], '--dry-run']
res = subprocess.run(resume_cmd, capture_output=True, text=True)
if res.returncode == 0:
s['last_visible_status'] = 'resume_verified'
else:
s['last_visible_status'] = f"resume dry-run failed: {res.stderr.strip() or res.stdout.strip()}"
# the completed handler's pin (Issue #40): own id, source and verified together
autopin.pin_row(s, agent, uuid)
s['last_visible_status'] = autopin.resume_dry_run(skills_dir, cwd, agent, s['name'])
id_name = 'session' if agent == 'claude' else 'conversation'
if not degraded:
@@ -783,314 +775,40 @@ for s in d.get('herdr_sessions', []):
else:
_pin_and_verify_resume(s, 'claude', cwd, uuid, degraded=True)
# === drift C (agy): agy 새 session id materialize (per-row own id) ===
# === drift C (agy), drift C (hermes), drift C (opencode), drift C (codex) ===
# Post-discovered ids (Issue #40): the adapter's capture_active_session_id
# decides - the same decision the completed handler (lib_py.agents.autopin)
# makes - so the two paths cannot drift apart. Exactly one verified candidate
# newer than the session that no other row records; agy additionally needs its
# presence lock held by this row's pane (its cwd cache alone proves nothing).
for s in d.get('herdr_sessions', []):
if row_agent(s) != 'agy':
agent = row_agent(s)
if agent not in ('agy', 'hermes', 'opencode', 'codex'):
continue
if s.get('status') != 'running':
continue
if s.get('agy_conversation_id_own'):
if s.get(OWN_KEY_BY_AGENT[agent]):
continue
cwd = (s.get('pane') or {}).get('cwd', '')
if not cwd:
continue
conv_dir = f"{home}/.gemini/antigravity-cli/conversations"
if not os.path.isdir(conv_dir):
continue
dbs = sorted(glob.glob(f"{conv_dir}/*.db"), key=os.path.getmtime, reverse=True)
sibling_claimed = [
other.get('agy_conversation_id_own')
for other in d.get('herdr_sessions', [])
if other is not s
and (other.get('pane') or {}).get('cwd') == cwd
and other.get('status') not in ('stopped', 'terminated')
and other.get('agy_conversation_id_own')
]
s_eval = dict(s)
s_eval['_sibling_claimed_uuids'] = sibling_claimed
valid_candidates = []
for db in dbs:
uuid = os.path.basename(db)[:-3]
if verify_session_uuid(cwd, 'agy', uuid, s_eval, mode="discover"):
valid_candidates.append(uuid)
if len(valid_candidates) > 1:
res = autopin.capture(s, agent, d, home_dir=home)
if res['status'] == 'ambiguous':
n = len(res['candidates'])
drifts.append({'class': 'C-ambiguous', 'name': s['name'],
'msg': f"{s['name']}: {len(valid_candidates)} candidate transcripts newer than session epoch; not pinning"})
s['last_visible_status'] = f"ambiguous: {len(valid_candidates)} candidates"
'msg': f"{s['name']}: {n} candidate transcripts newer than session epoch; not pinning"})
s['last_visible_status'] = f"ambiguous: {n} candidates"
actions.append(f"ambiguous candidates: {s['name']}")
if len(valid_candidates) == 1:
uuid = valid_candidates[0]
cmd = ['bash', '-c', f'source "{lib_sh}" && verify_tui_viewport "{s["name"]}" "agy" "{cwd}"']
elif res['status'] == 'captured':
uuid = res['session_id']
cmd = ['bash', '-c', f'source "{lib_sh}" && verify_tui_viewport "{s["name"]}" "{agent}" "{cwd}"']
rc = subprocess.run(cmd).returncode
if rc == 0:
_pin_and_verify_resume(s, 'agy', cwd, uuid, degraded=False)
_pin_and_verify_resume(s, agent, cwd, uuid, degraded=False)
elif rc == 1:
drifts.append({'class': 'C-warn', 'name': s['name'], 'msg': f"{s['name']}: TUI viewport mismatch for candidate {uuid} — not pinned, will retry next cycle"})
else:
_pin_and_verify_resume(s, 'agy', cwd, uuid, degraded=True)
# === drift C (hermes): hermes 새 session id materialize (per-row own id) ===
for s in d.get('herdr_sessions', []):
if row_agent(s) != 'hermes':
continue
if s.get('status') != 'running':
continue
if s.get('hermes_conversation_id_own'):
continue
cwd = (s.get('pane') or {}).get('cwd', '')
if not cwd:
continue
hdb = f"{home}/.hermes/state.db"
if not os.path.exists(hdb):
continue
epoch_threshold = s.get('herdr_session_epoch', 0)
sibling_claimed = [
other.get('hermes_conversation_id_own')
for other in d.get('herdr_sessions', [])
if other is not s
and (other.get('pane') or {}).get('cwd') == cwd
and other.get('status') not in ('stopped', 'terminated')
and other.get('hermes_conversation_id_own')
]
s_eval = dict(s)
s_eval['_sibling_claimed_uuids'] = sibling_claimed
valid_candidates = []
try:
conn = _ro_connect(hdb)
rows = conn.execute(
"SELECT id FROM sessions WHERE cwd=? AND started_at >= ? ORDER BY started_at DESC LIMIT 20",
(cwd, epoch_threshold)
).fetchall()
conn.close()
for (uuid,) in rows:
if uuid in (sibling_claimed or []):
continue
if verify_session_uuid(cwd, 'hermes', uuid, s_eval, mode="discover"):
valid_candidates.append(uuid)
except Exception:
pass
if len(valid_candidates) > 1:
drifts.append({'class': 'C-ambiguous', 'name': s['name'],
'msg': f"{s['name']}: {len(valid_candidates)} candidate transcripts newer than session epoch; not pinning"})
s['last_visible_status'] = f"ambiguous: {len(valid_candidates)} candidates"
actions.append(f"ambiguous candidates: {s['name']}")
if len(valid_candidates) == 1:
uuid = valid_candidates[0]
cmd = ['bash', '-c', f'source "{lib_sh}" && verify_tui_viewport "{s["name"]}" "hermes" "{cwd}"']
rc = subprocess.run(cmd).returncode
if rc == 0:
_pin_and_verify_resume(s, 'hermes', cwd, uuid, degraded=False)
elif rc == 1:
drifts.append({'class': 'C-warn', 'name': s['name'], 'msg': f"{s['name']}: TUI viewport mismatch for candidate {uuid} — not pinned, will retry next cycle"})
else:
_pin_and_verify_resume(s, 'hermes', cwd, uuid, degraded=True)
# === drift C (opencode): opencode 새 session id materialize (per-row own id) ===
for s in d.get('herdr_sessions', []):
if row_agent(s) != 'opencode':
continue
if s.get('status') != 'running':
continue
if s.get('opencode_session_id_own'):
continue
cwd = (s.get('pane') or {}).get('cwd', '')
if not cwd:
continue
# Resolve DB path via shared OpenCodeAgentAdapter helper with fallbacks
odb = None
try:
op_adapter = get_adapter('opencode')
if op_adapter:
odb = op_adapter._resolve_db(DiscoveryContext(workspace=cwd, agent_name='opencode', home_dir=home))
except Exception:
pass
if not odb:
opencode_base = f"{home}/.local/share/opencode"
candidates = [
f"{opencode_base}/opencode.db",
f"{opencode_base}/storage.db",
]
for cand in candidates:
if os.path.exists(cand):
odb = cand
break
if not odb and os.path.isdir(f"{opencode_base}/project"):
for root, _, files in os.walk(f"{opencode_base}/project"):
for f in files:
if f.endswith(".db"):
odb = os.path.join(root, f)
break
if odb and os.path.exists(odb):
break
if not odb or not os.path.exists(odb):
continue
epoch_threshold = s.get('herdr_session_epoch', 0)
sibling_claimed = [
other.get('opencode_session_id_own')
for other in d.get('herdr_sessions', [])
if other is not s
and (other.get('pane') or {}).get('cwd') == cwd
and other.get('status') not in ('stopped', 'terminated')
and other.get('opencode_session_id_own')
]
s_eval = dict(s)
s_eval['_sibling_claimed_uuids'] = sibling_claimed
valid_candidates = []
try:
# read-only: nothing is created or removed beside the agent's db (Issue #60)
conn = _ro_connect(odb)
cursor = conn.cursor()
# opencode 2.x keeps sessions in session_v2 (a migrated 1.x `session`
# table is no longer written).
cursor.execute("PRAGMA table_info(session_v2)")
cols = [col[1] for col in cursor.fetchall()]
if cols:
cwd_col = "directory" if "directory" in cols else ("cwd" if "cwd" in cols else None)
time_col = "time_created" if "time_created" in cols else ("created_at" if "created_at" in cols else ("started_at" if "started_at" in cols else None))
query = "SELECT id FROM session_v2"
params = []
where_clauses = []
if "parent_id" in cols:
# a subagent's session shares the directory and is newer
where_clauses.append("parent_id IS NULL")
if cwd_col and cwd:
where_clauses.append(f"{cwd_col} = ?")
params.append(cwd)
if time_col and epoch_threshold:
if time_col == "time_created":
where_clauses.append(f"{time_col} >= ?")
params.append(int(float(epoch_threshold) * 1000))
else:
where_clauses.append(f"{time_col} >= ?")
params.append(epoch_threshold)
if where_clauses:
query += " WHERE " + " AND ".join(where_clauses)
if time_col:
query += f" ORDER BY {time_col} DESC"
query += " LIMIT 20"
rows = cursor.execute(query, params).fetchall()
conn.close()
for (uuid,) in rows:
if uuid in (sibling_claimed or []):
continue
if verify_session_uuid(cwd, 'opencode', uuid, s_eval, mode="discover"):
valid_candidates.append(uuid)
else:
conn.close()
except Exception:
pass
if len(valid_candidates) > 1:
drifts.append({'class': 'C-ambiguous', 'name': s['name'],
'msg': f"{s['name']}: {len(valid_candidates)} candidate transcripts newer than session epoch; not pinning"})
s['last_visible_status'] = f"ambiguous: {len(valid_candidates)} candidates"
actions.append(f"ambiguous candidates: {s['name']}")
if len(valid_candidates) == 1:
uuid = valid_candidates[0]
cmd = ['bash', '-c', f'source "{lib_sh}" && verify_tui_viewport "{s["name"]}" "opencode" "{cwd}"']
rc = subprocess.run(cmd).returncode
if rc == 0:
_pin_and_verify_resume(s, 'opencode', cwd, uuid, degraded=False)
elif rc == 1:
drifts.append({'class': 'C-warn', 'name': s['name'], 'msg': f"{s['name']}: TUI viewport mismatch for candidate {uuid} — not pinned, will retry next cycle"})
else:
_pin_and_verify_resume(s, 'opencode', cwd, uuid, degraded=True)
# === drift C (codex): codex 새 session id materialize (per-row own id) ===
for s in d.get('herdr_sessions', []):
if row_agent(s) != 'codex':
continue
if s.get('status') != 'running':
continue
if s.get('codex_session_id_own'):
continue
cwd = (s.get('pane') or {}).get('cwd', '')
if not cwd:
continue
codex_home = os.environ.get('CODEX_HOME') or f"{home}/.codex"
epoch_threshold = s.get('herdr_session_epoch', 0)
sibling_claimed = [
other.get('codex_session_id_own')
for other in d.get('herdr_sessions', [])
if other is not s
and (other.get('pane') or {}).get('cwd') == cwd
and other.get('status') not in ('stopped', 'terminated')
and other.get('codex_session_id_own')
]
s_eval = dict(s)
s_eval['_sibling_claimed_uuids'] = sibling_claimed
valid_candidates = []
try:
# Rev.2: state DB first (indexed, single query) — dynamic filename,
# no hardcoded suffix (see §1.5 / §2 _state_db()).
state_matches = glob.glob(f"{codex_home}/state_*.sqlite")
state_db = max(state_matches, key=os.path.getmtime) if state_matches else None
rows = []
if state_db and os.path.exists(state_db):
try:
conn = _ro_connect(state_db)
cols = [c[1] for c in conn.execute("PRAGMA table_info(threads)").fetchall()]
order_col = "updated_at" if "updated_at" in cols else ("created_at" if "created_at" in cols else None)
if order_col:
rows = conn.execute(
f"SELECT id, {order_col} FROM threads WHERE cwd=? ORDER BY {order_col} DESC LIMIT 20",
(cwd,)
).fetchall()
else:
rows = [(r[0], int(time.time())) for r in conn.execute("SELECT id FROM threads WHERE cwd=? LIMIT 20", (cwd,)).fetchall()]
conn.close()
except Exception:
rows = []
if not rows:
# Fallback: bounded glob over the date-nested tree
paths = sorted(
glob.glob(f"{codex_home}/sessions/*/*/*/rollout-*.jsonl"),
key=os.path.getmtime, reverse=True
)[:50]
rows = [(os.path.basename(p)[:-6][-36:], os.path.getmtime(p)) for p in paths]
for uuid, ts in rows:
if epoch_threshold and ts and ts < epoch_threshold:
continue
if uuid in (sibling_claimed or []):
continue
if verify_session_uuid(cwd, 'codex', uuid, s_eval, mode="discover"):
valid_candidates.append(uuid)
except Exception:
pass
if len(valid_candidates) > 1:
drifts.append({'class': 'C-ambiguous', 'name': s['name'],
'msg': f"{s['name']}: {len(valid_candidates)} candidate transcripts newer than session epoch; not pinning"})
s['last_visible_status'] = f"ambiguous: {len(valid_candidates)} candidates"
actions.append(f"ambiguous candidates: {s['name']}")
elif len(valid_candidates) == 1:
uuid = valid_candidates[0]
cmd = ['bash', '-c', f'source "{lib_sh}" && verify_tui_viewport "{s["name"]}" "codex" "{cwd}"']
rc = subprocess.run(cmd).returncode
if rc == 0:
_pin_and_verify_resume(s, 'codex', cwd, uuid, degraded=False)
elif rc == 1:
drifts.append({'class': 'C-warn', 'name': s['name'], 'msg': f"{s['name']}: TUI viewport mismatch for candidate {uuid} — not pinned, will retry next cycle"})
else:
_pin_and_verify_resume(s, 'codex', cwd, uuid, degraded=True)
_pin_and_verify_resume(s, agent, cwd, uuid, degraded=True)
result = {
'timestamp': now_iso,