Jyotisha/docs/research/repo_cleanup_promotion_map_2026_07_01.md

Repo Cleanup And Promotion Map - 2026-07-01

Purpose

This note turns the current fragment-governance findings into one concrete cleanup map.

It does not delete historical assets. It defines:

1. what should remain reference-only,
2. what should be promoted into repo truth next,
3. what must never flow back into runtime truth.

This file is for practical day-to-day use during multi-window work.

Rule Summary

Three layers need active discipline:

1. docs/research/local_drafts/2026-06
2. external Gemini work-brain scratch under <home>/.gemini/antigravity-ide/brain
3. WorkBuddy mirror under <home>/.workbuddy/skills/jyotish-vedic-astrology

The repo runtime truth remains:

• SKILL.md
• AGENTS.md
• scripts/
• tests/
• references/
• canonical committed docs/research/*.md

A. Local Drafts 2026-06

Status:

• high-value source pool
• not runtime truth
• not directly callable by workflow

Current governance source:

• [local_drafts_2026_06_disposition.md](/docs/research/local_drafts_2026_06_disposition.md)

Promote First

These are the highest-value draft fronts to convert into canonical repo truth next:

1. antigravity_round36_tajika_sahams_external_closure_pack_2026_06_26.md
2. antigravity_round37_dasha_external_oracle_shortest_closure_board_2026_06_26.md
3. antigravity_round37_shadbala_absolute_value_frontier_board_2026_06_26.md
4. antigravity_round39_yogi_wealth_bridge_audit_2026_06_28.md
5. three_fronts_skill_depth_audit_2026_06_26.md
6. current_skill_core_gap_rerank_2026_06_26.md
7. five_hard_fronts_master_board_2026_06_26.md

Promotion Targets

When promoted, they should end up in one of four canonical forms only:

• docs/research/*.md
• references/*.md
• tests/*.py
• scripts/*.py

Current promoted packs:

• [promote_first_repo_truth_pack_2026_07_01.md](/docs/research/promote_first_repo_truth_pack_2026_07_01.md)
• [promote_second_repo_truth_pack_2026_07_01.md](/docs/research/promote_second_repo_truth_pack_2026_07_01.md)
• [promote_third_repo_truth_pack_2026_07_01.md](/docs/research/promote_third_repo_truth_pack_2026_07_01.md)
• [promote_fourth_repo_truth_pack_2026_07_01.md](/docs/research/promote_fourth_repo_truth_pack_2026_07_01.md)

Do Not Do

• do not import draft code directly into runtime
• do not cite drafts as final capability truth without re-anchoring
• do not delete the draft pack just to reduce clutter

B. External Gemini Work-Brain

Status:

• recovery-only
• useful for archaeology
• not repo truth

Examples already flagged by preflight:

• run_vedastro.py
• test_vedastro_events.py
• test_vedastro_network.py
• vedastro_capability_gap_analysis.md

Allowed Use

Allowed:

• compare ideas
• recover lost exploration
• re-anchor findings into repo docs or code

Not allowed:

• treat these files as current implementation truth
• link runtime imports to them
• quote them as final project status without repo confirmation

Cleanup Action

The cleanup action here is documentation and triage, not repo-side deletion:

• if a work-brain artifact is still valuable, rewrite its claim into repo truth
• if it is obsolete, leave it in recovery-only status and stop re-reading it during normal implementation

C. WorkBuddy Mirror

Status:

• distribution mirror
• historical reference only
• explicitly not runtime source-of-truth

Canonical boundary sources:

• [unique_main_chain_map_2026_07_01.md](/docs/research/unique_main_chain_map_2026_07_01.md)
• [whole_project_fragment_sweep_and_vedastro_ledger_link_2026_06_28.md](/docs/research/whole_project_fragment_sweep_and_vedastro_ledger_link_2026_06_28.md)

Allowed Use

Allowed:

• parity check
• drift detection
• distribution packaging reference

Not allowed:

• reverse-copy runtime code into repo truth
• add mirror path to import path
• treat mirror docs as more current than the main repo

Required Pre-Work Checklist

Before touching:

• strict workflow
• VedAstro official ingestion
• benchmark/oracle closure
• wealth/career/relationship adjudicators
• fragment reuse claims

do this:

1. check canonical repo truth first
2. check local_drafts_2026_06_disposition.md
3. check preflight_fragment_scan.py output
4. only then inspect Gemini brain or WorkBuddy mirror if the canonical layer is insufficient

Current Best Cleanup Strategy

This is the shortest safe path:

1. keep runtime code untouched unless a real gap is proven
2. keep drafts in place but promote the top-value seven items above
3. keep Gemini work-brain strictly as recovery-only
4. keep WorkBuddy strictly as mirror/reference-only
5. enforce the boundary through tests and preflight output, not by memory

Boundary

This cleanup map exists to reduce repeated rediscovery cost.

It does not mean:

• local drafts are garbage
• external work-brain is useless
• WorkBuddy mirror should be deleted

It means all three must stay outside runtime truth until explicitly re-anchored.