[OPS] Onboard and refresh the approved company repository fleet

[100yenadmin]

Parent: #130

Goal

Create one approved repository manifest, refresh stale approved indexes, and onboard every eligible active company repository once with Voyage embeddings.

Policy

Include non-archived electricsheephq/* and 100yenadmin/* repositories except test/canary/quarantine/local-only and eva-workspace. Keep upstream OpenClaw and Hermes Agent as named exceptions. Replace third-party Hermes-LCM with 100yenadmin/hermes-lcm; remove Kitchenasty.

The portable manifest drives a generated local absolute-path manifest, Codex and both Claude allowlists/default, registry expectations, and the Voyage premium allowlist.

Acceptance and proof

Registry rows equal unique normalized remotes in the approved manifest; aliases are case-insensitively unique; every indexable repository has Voyage code-3 2,048-dimensional embeddings and live vector-index retrieval; true zero-text repositories are explicitly not-indexable. Refresh only through the coordinator after the merged source batch.

Evidence: /Volumes/LEXAR/Codex/session-notes/2026-07-20/gitnexus-repository-health-sprint/registry/.

Stop conditions

Stop on dirty/detached/ahead canonical checkout, analyzer conflict, failed Voyage readiness, provider/config mismatch, false embedding success, or a candidate outside approved ownership/exception policy. Voyage runs one repository at a time and has no host memory/swap admission gate; local embedding backends are outside this lane.

Exact next action

Merge operating-kit PR #59 and the GitNexus .3 source/release lanes, install the exact artifact, then render the approved manifest before queueing Voyage repositories sequentially.

[evaos-code-review-bot]

evaOS issue enrichment

Issue: #135 - [OPS] Onboard and refresh the approved company repository fleet
State: open; milestone: GitNexus Repository Health & Large-Repo Reliability Sprint

Related issues/PRs: #130, #59.
Existing labels: P1, evidence-gate.
Suggested labels: tests.
Suggested owners: none.

Related context

• EPIC: Restore truthful semantic retrieval and canonical index lifecycle #130 - mentioned in issue metadata; inspect for dependency, duplicate, or prior-decision risk.
• R20: Repin reconciliation from RC19 to stable v1.6.10 #59 - mentioned in issue metadata; inspect for dependency, duplicate, or prior-decision risk.

Agent-start packet

Problem shape: [OPS] Onboard and refresh the approved company repository fleet - Parent: #130 ## Goal Create one approved repository manifest, refresh stale approved indexes, and onboard every eligible active company repository once with Voyage embeddings. ## Policy Include non-archived
Product fit: Architecture/integration issue; check repo policy, prior decisions, and operational constraints before implementation.
Implementation wedge: Start with a bounded adapter/contract or dry-run packet before changing live runtime behavior.

Build / borrow / buy scan:

• Build: identify the smallest implementation wedge that proves the issue without broad rollout.
• Borrow: inspect same-repo prior art, repo memory, GitNexus context, and allowlisted cross-repo examples first.
• Buy/use: deferred; no product, architecture, UX, library-choice, integration, roadmap, or market-positioning trigger detected.

Candidate sources:

• VISION.md / repo policy / repo-memory.md / GitNexus packet.
• Same-repo issues and PRs referenced by issue metadata.

Acceptance criteria:

• Preserve and execute the issue's stated acceptance criteria.
• Include source citations for any external OSS/library/current-market claims.
• Keep issue-enrichment comments sticky and idempotent.
• Keep labels, owners, reviewers, and roadmap fields suggestion-only.

Proof plan:

• Focused unit/fixture test for the changed planner behavior.
• Dry-run issue-enrichment evidence packet before live comment posting.
• Redaction check for issue body, citations, and generated markdown.
• Contract or degraded-mode proof before runtime promotion.

Known traps:

• Do not treat related links as proof by themselves; each source needs a reason why it matters.
• Do not scan old issue backlogs or exceed repo-level enrichment throttles.

Non-goals:

• Do not auto-apply labels, owners, reviewers, roadmap fields, or milestones.
• Do not bulk-enrich old backlog issues from this planner output.
• Do not claim external market or OSS research was performed unless cited sources are present.

Context-source taxonomy:

• [enabled] vision_repo_policy_memory_gitnexus: Use repo vision, policy, memory, and GitNexus first before external research.
• [enabled] same_repo_issues_prs: Issue metadata contains same-repo references.
• [deferred] allowlisted_cross_repo_github: Deferred until product, UX, market-positioning, or library-choice signal appears.
• [deferred] external_oss_examples: Deferred for low-research-value issue class.
• [deferred] library_api_docs: Deferred unless a library/API decision is present.
• [deferred] current_market_examples: Deferred for implementation-only issues.
• [enabled] internal_mcp_docs: Use internal docs/MCP context when available; missing context is degraded, not blocking.

Validation suggestions

• Confirm owner, acceptance criteria, and validation evidence before implementation.

Triage gaps

• Owner or reviewer signal not detected in issue body.

No labels, owners, reviewers, or roadmap fields were changed by this bot.

[100yenadmin]

Fleet checkout preparation update (no indexing or registry mutation):

• All 67 active canonical paths now exist.
• All 67 origin URLs normalize to the approved manifest remote.
• The 42 previously missing approved-root checkouts were cloned strictly one at a time; all are clean at verified 40-character HEADs; zero clone failures and zero leftover temporary clone directories.
• hermes-lcm was cleanly fast-forwarded from its preserved upstream HEAD to 100yenadmin/hermes-lcm/main (+183), with the old remote retained as upstream.
• codex was cleanly fast-forwarded to 100yenadmin/codex/main (+5), with OpenAI retained as upstream.
• Four pre-existing dirty canonical roots remain untouched and explicit: evaOS-GUI, evaos-cortex, WorldOS, and agent-skill-debloater.

Voyage indexing remains gated only on the reviewed .3 source/artifact and will run one repository at a time. There is no memory/swap admission gate.

[100yenadmin]

Read-only canonical-manifest comparison is complete; no live registry/index/config mutation has started.

• Rendered approved fleet: 67 active rows, 67 unique normalized remotes, 67 case-insensitive aliases
• Checkout verification: 67/67 paths present, 0 origin mismatches
• Current registry: 50 rows, 37 normalized remotes, 39 aliases
• Gap: 36 approved remotes not yet registered, 9 canonical-path corrections, 6 extra rows
• Duplicates: 5 normalized-remote collision families and 4 alias-collision families
• Live registry remains byte-identical to the captured baseline (sha256 a29f141cc52a433c6b100d26e1fafca7cf7bd50b17c8f8c8109996c1e86a0aea)

Voyage onboarding will run one repository at a time and has no host memory/swap/RSS admission gate. Registry cleanup and indexing remain blocked on the exact .3 release/install and read-only .3 doctor proof.

[100yenadmin]

Fleet execution update: Voyage indexing now uses at most two background slots, with no host memory/swap gate. Jobs must target different normalized remotes and index paths, use staged mode, and keep independent evidence; final registry mutation/cleanup remains serialized. Local embeddings remain single-repo and resource-gated, and genuinely large repos remain solo unless paired only with a clearly small repo. Completed new semantic indexes include longevity-protocol-compass, eva-skills, backboard-openclaw, backboard-locomo-benchmark, locomo, honcho, and evermemos, each with count agreement and live vector-index proof. Current slots: memobase and zep-memory. The remaining brand-new onboarding lane is non-blocking for the .3 release/runtime proof.

[100yenadmin]

Fleet queue update under the approved two-slot Voyage policy:

• Active: company-fork hermes-lcm candidate and ladybug, on distinct remotes/index paths with staged mode and separate evidence logs.
• Held: openviking and clawsweeper. Both initial runs stopped on the same transient Voyage 30-second timeout; their single bounded .3 retries then reproduced the restored-embedding duplicate-primary-key defect fixed by PR fix: delete restored embeddings by exact identity #156. Neither canonical index was promoted, and both staged generations are preserved.
• No memory, swap, RSS, or free-memory gate was used. No third analyzer is running.

The fleet lane continues on clean first-pass repositories. The two interrupted repositories wait for the installed patch containing #156; no further .3 retry will be made.