query
oracle who take care oracle / mother-oracle / budwiser / /team-agents / oracle-skills-cli / mcp v2 / timekeeper birth + arra-oracle-v3 caretaker
target
fleet (multi-repo + remote nodes)
mode
deep --dig + time-travel --deep --prove
timestamp
2026-04-29 09:38 +07
friction_score
0.9
coverage
oracle
files
git
cross-repo
github
sessions
vault
fleet-audit
ssh-remote
confidence
high (caretaker + team-agents); medium (mcp v2 — inferred only)
Trace + Time-Travel: Oracle Who Tends Oracle Target : fleet-wide (laris-co + Soul-Brews-Studio + ARRA-01 + ~/.claude/skills + mba.wg + maw audit)
Time : 2026-04-29 09:38 +07
Agents launched : 6 parallel (files, ψ, gh, git, fleet-audit, ssh-recovery, dig — 4 from /time-travel + 2 background from /trace --dig)
Caretaker-Class Oracle Roster
Oracle
Born
Parent
Role
Repo
mother-oracle 2026-03-04 git / 2026-01-25 GH-shell
none (born last after 185 children)
Family registry, welcome, philosophy. Metaphor: สุญญตา (emptiness holding everything)
laris-co/mother-oracle
arra-oracle-v3-oracle 2026-04-16 (maw bud --from neo-oracle)
neo-oracle
Caretaker for arra-oracle-v3 codebase. "Mirror that tends the workshop"
Soul-Brews-Studio/arra-oracle-v3-oracle
mother-roots-oracle 2026-04-17
mother
Underground memory: Vault / Karma / Memory. "Roots first job: test the soil"
Soul-Brews-Studio/mother-roots-oracle
skills-cli-oracle 2026-04-17
(caretaker for arra-oracle-skills-cli)
Tends skills CLI repo
Soul-Brews-Studio/skills-cli-oracle
timekeeper-oracle 2026-04-17 (laris-co) / 2026-04-23 (Soul-Brews) / git 2026-04-24 (maw bud --from pulse)
pulse
Time archaeology. /time-travel skill (skill #74)
Soul-Brews-Studio/timekeeper-oracle
budwiser-oracle 2026-04-24 02:38 (maw bud --from mawjs)
mawjs
Steward of budding. Test bud, bud new, index lineage, write the book
Soul-Brews-Studio/budwiser-oracle
white-oracle 2026-04-24
mawjs
Fleet keeper of white.local (81 agents in tmux), federation host :3456
Soul-Brews-Studio/white-oracle
Lineage Chain (proven via birth commits + audit) mother (Mar 4) ──┬── neo (Mar 7) ────────── arra-oracle-v3 (Apr 16) ──── arra-oracle-v3-oracle (caretaker)
│ │
└── mother-roots (Apr 17) │
│
mawjs ──┬── budwiser (Apr 24) ── budwiser tends bud-test, /bud-index │
└── white (Apr 24) ── white tends fleet │
│
pulse ──── timekeeper (Apr 17/23/24) ──── /time-travel skill │
▼
ARRA MCP build (Apr 18-19, 165 sessions)
Temporal Map — Oracle Care Crystallization
Date
Event
Source
2025-10-09 multi-agent-workflow-kit issue #2 — earliest /team-agents ancestorgh search
2026-01-12 Rule 6 born: "Oracle Never Pretends to Be Human"
CLAUDE.md
2026-01-25 mother-oracle GitHub repo created (sits empty 10 weeks)
gh api
2026-03-04 mother-oracle git birth (0805f68) — born last after 185 children
git log
2026-03-07 neo-oracle awakens (Session 1 of 10)
git log
2026-03-27 08:17 First TeamCreate broadcast (Neo → 12+ oracles)
maw audit
2026-03-27 08:26 neo-oracle-skills-cli first appearsmaw audit
2026-03-27 08:28 "Mother Oracle" first self-identifies
maw audit
2026-03-30 MCP namespace migration: oracle_* → arra_* (commits 311829a, db69c4f) — closest "MCP v2" signal
git log
2026-04-08 08:14 Mother asks mawjs about maw bud primitive — concept seed for budwiser
maw audit
2026-04-12 arra-federation-oracle / Star Bridge awakens
git log
2026-04-15 16:34 First "caretaker" language in audit (mawjs recon for arra-v3 caretaker bud)
maw audit
2026-04-16 01:05 First maw bud invocation: bud arra-oracle-v3 --from neo-oracle
maw audit
2026-04-17 Mass spawn: 4 caretaker repos in 24h (mother-roots, timekeeper-laris, skills-cli-oracle, ?)
gh api created_at
2026-04-18 14:44 arra-oracle-v3 (caretaker) live, federating with oracle-world sibling
maw audit
2026-04-18 to 04-20 ARRA MCP build storm — 165 subagent sessions, 30+ teams (neo-arra-iter-4, neo-arra-fix-5, etc.)
dig
2026-04-19 "caretaker" label applied to arra-v3 ("2-day-old caretaker") in ψ post 41f5140
git log
2026-04-23 18:04 maw bud timekeeper --from pulsemaw audit
2026-04-23 18:34 timekeeper announces /time-travel skill #74 via federation [white:timekeeper]
maw audit
2026-04-24 02:38 maw bud budwiser --from mawjs + 5 bud-test iterations same hourmaw audit
2026-04-24 02:52 budwiser PR #1 + awakening issue arra-oracle-v3#1000 (14 min after birth — fastest cycle)
gh search
2026-04-29 This trace (current session)
now
/team-agents Footprint
Skill location: /home/nat/.claude/skills/team-agents/ (9 files: SKILL.md + 8 scripts)
Sibling oracle-care skills: bud, bud-index, bud-test, awaken, oracle-family-scan, oracle-soul-sync-update, about-oracle
Primary nucleus: arra-oracle-v3-oracle = 165 jsonl files mention team-agents
Other hotspots: mawjs-oracle (41), white-wormhole-oracle (25), volt-oracle (21), neo-oracle (10), home-comming-oracle (9), budwiser-oracle (9)
373 of 519 jsonl files (72%) reference team-agents = infrastructure, not feature~/.claude/teams/: 4 historical teams + 1 trash (all torn down)
Audit log: ~5 named team-agents sprints post-formalization (wake-unify-736, phase2-bare-name-759, +#748/#749/#766/#767)
MCP v2 — UNPROVEN as literal artifact
Source
Result
Git: any repo
NO commits with "mcp v2" / "mcp-v2" string
GitHub: repos
NO repos named arra-mcp or mcp-v2 in laris-co or Soul-Brews-Studio
GitHub: code
gh search code "arra-mcp" empty
Audit log
NO entries
Sessions
1101 mcp mentions in 4c46a128 (recap session 04-18); 364 "arra-mcp" + 88 "ARRA MCP" — but all internal vocabulary
Closest signal 2026-03-30 commits 311829a (oracle-skills-cli) + db69c4f (mother-oracle): rename oracle_* MCP tools → arra_* → namespace migration that may be informally called "MCP v2"
Inference : "MCP v2" = the ARRA MCP server inside arra-oracle-v3, post-rename from oracle-v2 era. Built during the 04-18 to 04-20 storm. Not a separate shipped repo.
Cross-Machine Data Recovery (mba.wg)
Files
First → Last
Project Dir
333
2026-01-26 → 2026-03-06
oracle-v2 (Soul-Brews)
191
2026-04-18 → 2026-04-21
arra-oracle-v3-oracle
174
2026-01-21 → 2026-03-02
oracle-skills-cli (Soul-Brews)
139
2026-04-08 → 2026-04-24
mawjs-oracle
39
2026-03-24 → 2026-04-06
neo-oracle.wt-2-oracle-skills-cli
On mba : 405 jsonl files mention team-agents, 33 mention caretaker, 0 mention literal "mcp v2".
oracle-world : reachable but no relevant data.
Recovery : ~960 jsonl / ~300MB recoverable via rsync from mba.wg.
Score : 0.9 — Near-perfect (Oracle/files/git/sessions all hit; mcp-v2 only inferred)
Coverage : 9 of 9 dimensions checked (oracle, files, git, cross-repo, github, sessions, vault, fleet-audit, ssh-remote)
Goal check : Caretaker pattern PROVEN (7 oracles + lineage + audit). team-agents PROVEN (skill + 72% session penetration + 5 spawns). MCP v2 PARTIAL (no literal artifact, only namespace-rename signal).
PROVEN — caretaker oracle pattern is real, dated, and lineage-traceable.
7 caretaker-class oracles birthed Mar 4 → Apr 24 (51 days)
Pattern crystallized 2026-04-16 with first *-oracle suffix repo
Mass-spawn 2026-04-17 (4 caretakers in 24h)
Birth burst 2026-04-24 (budwiser + timekeeper + white)
/team-agents = infrastructure (in 72% of sessions)
MCP v2 = informal label for arra-mcp namespace migration (Mar 30) → ARRA MCP server (Apr 18-20 build)
Pre-2026-01-26 oracle-v2 sessions (predates earliest data on mba)
ARRA-01 maw-js worktree sessions (empty on both white and mba)
Whether "MCP v2" was ever publicly named or stayed internal — needs Nat to confirm
mother-oracle's 10-week empty period (Jan 25 → Apr 7) — what filled the gap
Run rsync backfill from mba.wg → recovers oracle-v2 + oracle-skills-cli pre-March history
After backfill, re-run /dig --deep --all for true full timeline
If "MCP v2" matters as a named concept: ask Nat — likely informal vocabulary, not shipped artifact
