Skip to content

docs: ecosystem audit + tiered eliminate/archive/keep plan#2

Open
HeadyMe wants to merge 5 commits into
mainfrom
claude/ecosystem-audit-gfl0f4
Open

docs: ecosystem audit + tiered eliminate/archive/keep plan#2
HeadyMe wants to merge 5 commits into
mainfrom
claude/ecosystem-audit-gfl0f4

Conversation

@HeadyMe

@HeadyMe HeadyMe commented Jun 19, 2026

Copy link
Copy Markdown

Summary

A read-only audit of the Heady ecosystem with a tiered, gated lifecycle plan to classify every repo/service/file as keep / archive / eliminate. No irreversible action taken — this is the plan for owner review.

What was inspected

  • GitHub: 114 repos (19 HeadySystems + 95 HeadyAI).
  • Cloudflare: 37 live Workers.
  • Local: the 4 cloned repos (deep), incl. heady-ai legacy-vs-rebuild divergence (320 commits on main not in rebuild).

Key findings

  • Canonical line = HeadySystems/heady-ai @ rebuild (default branch); main is legacy.
  • Large legacy strata: archived battle-arena heady-rebuild-* repos, pre-rebuild monorepos, and 7+ duplicate Heady-Main/Heady-Staging/Heady-Testing clones (incl. hash-suffixed forks).
  • 37 Workers contain prod/non-prod duplicates and 4 overlapping routers to consolidate.

Honest constraints (in the doc, §0)

  • Only 4 repos are in write-scope; the other ~110 are read-only metadata — I can't archive them here (list_repos/add_repo unavailable). Org-wide execution needs expanded access or an admin running emitted scripts.
  • Destructive steps are gated; the correct execution engine is the rebuild's own coherence/awareness/projection-lifecycle tooling, not ad-hoc deletion.
  • External MCPs (Stripe/Slack/HF/…) deliberately not sprayed — no legacy-vs-rebuild signal there. /heady-omni-sync etc. not fired (need the heady-manager runtime; firing a sync blindly is destructive-risk).

Decisions requested (doc §7)

  1. HeadyAutoContext vs heady-context — which is canonical?
  2. headydocs vs heady-docs — consolidate to which?
  3. Confirm heady-production history is migrated to rebuild before archiving it.
  4. Approve Tier A (run scanners + commit the file-level manifest) and route Tier C access.

🤖 Generated with Claude Code

https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ


Generated by Claude Code

HeadyConnection and others added 5 commits June 19, 2026 13:02
Read-only audit across 114 repos (19 HeadySystems + 95 HeadyAI), 37 Cloudflare
Workers, and the 4 cloned repos. Identifies the canonical line (heady-ai@rebuild;
main is legacy with 320 unique commits), classifies repos/workers as
keep/archive/eliminate, and lays out a tiered execution plan whose destructive
steps are gated on owner confirmation and (for ~110 out-of-scope repos) expanded
access. Execution engine = the rebuild's own coherence/awareness/lifecycle
tooling, not ad-hoc deletion.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ
Execute the approved Both-A+C pass:
- Tier A: ran the canonical repo's scanners (coherence PASS, 0 contradictions,
  269-node map / 370-var registry). knip absent + legacy scan:stale broken on
  rebuild (recorded as gaps). Static pass: rebuild is lean — only 7 intentional
  governance/legacy files, patent IP blobs (LFS candidates), 42 dup-hash groups
  to review. Conclusion: reclamation is at org level, not in the canonical repo.
- Tier C: scripts/ecosystem-cleanup.sh — dry-run by default, archives 12+1
  duplicate/legacy repos (incl. HeadyAutoContext per owner: rebuild is canonical
  context SoT), double-gates 18 deletions, retires 6 duplicate Workers, stages
  the heady-ai/main freeze. Requires an admin (org repos out of session scope).
- Recorded owner decisions: canonical context = heady-ai@rebuild; execution = A+C.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ
…don't archive

Owner decision reversal: the context repos are curated corpora (knowledge,
patents, research, prompts for AI/NotebookLM/onboarding), distinct from the
rebuild monorepo's runtime SoT — not redundant. Remove HeadyAutoContext from the
Tier C archive list; both repos are KEEP. Boundary documented in ECOSYSTEM_AUDIT.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ
Document why the 9 archived heady-rebuild-* repos are safe to eliminate: per
battle-synthesis-report.json they were a 9-way model rebuild contest, synthesized
not winner-take-all. Each candidate won a different axis; the best component of
each (Claude DI, GPT-5.4 registry, Gemini module-graph, Perplexity security,
Codex TDD, HeadyCoder router) was folded into rebuild. Winner/loser is
angle-dependent (the CSL cosine principle); the repos are losers of redundancy,
not value — their winning projection already lives in rebuild.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ
The earlier draft said the canonical repo had nothing to eliminate; that was
before knip was wired (PR #228) and couldn't see dead files. The knip ∩
dependency-cruiser gate surfaced 8 confirmed dead files (advisory, gated).
Recorded per the standing rule that stale docs are a defect.

Co-Authored-By: Claude Opus 4.8 (1M context) <[email protected]>
Claude-Session: https://claude.ai/code/session_016WqBXQGJkjhG9RdAFiL2MZ
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants