chore(catalog): seed 14 confident repos + queue 37 for triage#33
Conversation
After the discovery-canonicalization fix de-noised the catalog gap (79→58 →54 real), seed the 14 medium-confidence repos as 'maintain' (owner/lifecycle/ cadence/disposition) — catalog-flagged drops 54→40. The 40 remaining are low-confidence (auditor can't derive disposition); the 37 real ones (3 junk excluded) are captured in docs/catalog-triage-worklist-2026-06-02.md grouped by activity for operator batch-triage.
![chatgpt-codex-connector[bot]](https://avatars.githubusercontent.com/in/1144995?s=60&v=4){kind=small}
There was a problem hiding this comment.
💡 Codex Review
Here are some automated review suggestions for this pull request.
Reviewed commit: b2d3495d39
ℹ️ About Codex in GitHub
Your team has set up Codex to review pull requests in this repo. Reviews are triggered when you
- Open a pull request for review
- Mark a draft as ready
- Comment "@codex review".
If Codex has suggestions, it will comment; otherwise it will react with 👍.
Codex can also answer questions or update the PR. Try commenting "@codex address that feedback".
| review_cadence: monthly | ||
| intended_disposition: maintain | ||
| category: it-work | ||
| CryptForge: |
There was a problem hiding this comment.
For the newly seeded repos that live under Fun:GamePrjs/ or ITPRJsViaClaude/ (e.g. CryptForge, DeepTank, DesktopPEt, EarthPulse, SmartClipboard, SnippetLibrary, TicketDashboard as shown in output/doc-truth-up-full-2026-05-30.log), leaving category unset does not fall through to the path group: catalog repo entries are normalized with the global default first, and _select_declared() then picks that repo value before the group value. These repos will therefore be reported as vanity (and Fun:GamePrjs tool provenance as unknown) instead of their path-derived group classification, skewing catalog summaries and downstream portfolio truth data.
Useful? React with 👍 / 👎.
|
|
||
| 37 repos flagged `catalog` (missing from `config/portfolio-catalog.yaml`) with **low** path-confidence — the auditor can't auto-derive a disposition, so each needs an operator call. | ||
|
|
||
| For each, decide `intended_disposition` (maintain | finish | archive | experiment | investigate) and `lifecycle_state`, then add to the catalog. The 14 medium-confidence repos were already auto-seeded as `maintain`. |
There was a problem hiding this comment.
This worklist tells operators to add investigate as an intended_disposition, but the catalog loader only accepts maintain, finish, archive, and experiment (VALID_INTENDED_DISPOSITIONS in src/portfolio_catalog.py). If an operator follows this instruction, the catalog entry is flagged invalid and the disposition is normalized away, so the repo remains without a stable operating path rather than getting an investigate outcome.
Useful? React with 👍 / 👎.
1 participant
Follows the discovery-canonicalization fix (#32). Seeds the 14 medium-confidence uncatalogued repos as maintain; catalog-flagged drops 54→40. The 37 low-confidence real repos are captured in docs/catalog-triage-worklist-2026-06-02.md for operator batch-triage.