Abandoned PRs — fuzzywigg/PikoClaw → nft2-me/PikoClaw Migration¶
Date: 2026-04-18
Scope: 12 of 37 open PRs in fuzzywigg/PikoClaw classified ABANDON (do not port to fresh nft2-me/PikoClaw repo). Final reconciled totals: 21 SALVAGE · 3 MAYBE · 12 ABANDON · 1 DUPE.
Reclassification note (2026-04-18, post full-title review): #190 (synthetic Slack export) and #198 (synthetic MBOX 7k) were initially tagged Duplicate/ABANDON on the assumption they mirrored #168–170. Full title review revealed distinct formats (Slack and MBOX vs CSV/EML), fitting the user's stated salvage rule for "other formats where little synthetic data is available." Both flipped to SALVAGE. Original ABANDON count was 14; final is 12.
Context¶
- Migration from personal private repo (GHAS unavailable, CodeQL
upload: neverconstraint) tonft2-meGitHub Team org. - Fresh-repo strategy, not git transfer. Old repo stays archived read-only; branches persist.
- "ABANDON" = do not cherry-pick commits into fresh repo. Not "delete" — branches remain for forensic/archaeology reference.
- This document records the disposition AND decides where each abandoned idea lives on in the nft2-me ecosystem, so nothing worth keeping vanishes with the PR close.
Abandonment Taxonomy¶
- Research-tier (L2 / L3 / L5): Valid ideas; too large or speculative for current phase. Preserve as issues in fresh repo or ADRs in
docs/adr/. - WIP-vague: No recoverable scope, unclear motivating problem. Close without follow-up.
- Duplicate: Another SALVAGE PR already covers the same ground. Redundant.
Per-PR Dispositions¶
| PR | Scope | Class | Reason | Preserved As | Imprint Location |
|---|---|---|---|---|---|
| #184 | L2 connector research | Research-tier | Needs design doc before code; too early | Issue: research/l2-connectors |
nft2-me/PikoClaw issues |
| #185 | L2 connector research | Research-tier | Overlaps #184 scope | Folded into #184 issue | nft2-me/PikoClaw issues |
| #186 | L2 connector research | Research-tier | Overlaps #184 scope | Folded into #184 issue | nft2-me/PikoClaw issues |
| #187 | Async pipeline WIP | WIP-vague | Large unfinished pivot; direction unclear | ADR: "Async pipeline — deferred" | nft2-me/PikoClaw/docs/adr/ |
| #188 | L3/L5 research | Research-tier | Speculative, no motivating problem yet | Collapsed research issue | nft2-me/PikoClaw issues |
| #189 | L3/L5 research | Research-tier | Same | Collapsed into parent research issue | nft2-me/PikoClaw issues |
| #192 | L3/L5 research | Research-tier | Speculative | Collapsed research issue | nft2-me/PikoClaw issues |
| #193 | L3/L5 research | Research-tier | Speculative | Collapsed | nft2-me/PikoClaw issues |
| #194 | L3/L5 research | Research-tier | Speculative | Collapsed | nft2-me/PikoClaw issues |
| #195 | WIP vague | WIP-vague | No recoverable signal | — | — |
| #196 | L3/L5 research | Research-tier | Speculative | Collapsed | nft2-me/PikoClaw issues |
| #197 | WIP vague | WIP-vague | No recoverable signal | — | — |
Imprint Surfaces — Where Ideas Live On¶
Four surfaces absorb the ideas that deserve to survive their abandoned PRs. Each future autonomous loop in the nft2-me ecosystem should consult these surfaces before opening new work in the affected areas.
nft2-me/PikoClawissues — Research ideas with labelsresearch/l2-connectors,research/l3-core,research/l5-semantic. Each issue carries: problem statement + link to oldfuzzywiggbranch SHA for code archaeology.nft2-me/PikoClaw/docs/adr/— Architectural decisions not to build X right now live as ADRs so the rationale survives context loss. Example:0003-async-pipeline-deferred.md.nft2-me/project-template/skills/— Patterns that generalize across projects (not PikoClaw-specific) become reusable skills. Example:batch-pr-governance.md(see "Lesson Imprinted" below).- Notion research backlog — Pure ideation not yet ready for an issue. Dated parking lot; graduates to GitHub issues when a motivating problem appears.
Ecosystem Imprint Plan — Execution Sequence¶
During migration execution, the following steps ensure nothing worth keeping is lost:
- Before closing ABANDON PRs in
fuzzywigg/PikoClaw: Extract titles + bodies into a staging markdown file (migration-staging.md, local only, not committed). - In fresh
nft2-me/PikoClaw: Create research issues per the "Preserved As" column. Apply labels. Link each issue to the old branch SHA. - Write
docs/adr/0003-async-pipeline-deferred.mdinto the fresh repo's initial commits, based on #187's direction. - Write
skills/batch-pr-governance.mdintonft2-me/project-templatewith the lesson below. - Close old PRs with the template comment (below), pointing back to this document and to any preserved issue/ADR.
Lesson Imprinted on the Dev Lifecycle¶
Finding: On 2026-04-12 the Copilot SWE Agent opened 33 PRs against fuzzywigg/PikoClaw/develop in a single day. This batch-dump overwhelmed review capacity, created CI-failure cascades from stale bases, and produced the exact backlog this migration is resolving.
Rule added to autonomy playbook (destination: nft2-me/project-template/skills/batch-pr-governance.md):
A single autonomous agent identity must not open more than N PRs per day against the same base branch without a human sign-off on at least one merge. Default N = 3. The first PR of a batch must reach green-merged state before the agent queues a second batch against the same base. Override requires explicit human
/unblockin the agent's task queue.
Why this rule: It prevents the failure mode where one agent opens so many PRs that none of them can be reviewed, and the backlog itself becomes the blocker. It also guarantees that CI stays validated against the current base rather than diverging across a cohort of stale branches.
Old-Repo Close-out Comment Template¶
Applied uniformly to all 37 closed PRs in fuzzywigg/PikoClaw regardless of disposition. The document linked here is the authoritative disposition record.
Closing as part of migration from
fuzzywigg/PikoClaw→nft2-me/PikoClaw(2026-04-18). Branch preserved in this repo (now archived read-only); disposition recorded in migration-abandoned-prs.md. If this PR's work was salvaged, it appears as a fresh PR in the new repo. If abandoned, the rationale and any preserved form (issue, ADR, skill) is listed there.
(URL above is a placeholder until the fresh repo exists and this doc is committed.)
Provenance & Revision¶
- Author: Autonomous session 2026-04-18, fuzzywigg (Andrew Pappas) directing
- Basis: GraphQL enumeration of 37 open PRs in fuzzywigg/PikoClaw + titles/authors/dates + human decision on flip-calls (#168–170 SALVAGE, #177 SALVAGE-with-polish)
- Superseded by: Any future rev of this doc inside
nft2-me/PikoClaw/docs/migration/ - Not authoritative for: SALVAGE dispositions — those live in the Notion State page and will be committed as
migration-salvage-inventory.mdalongside this doc during the migration execution commit