27 open PRs contain ~9 duplicate/superseded submissions — needs triage before merging

[dmchaledev]

Problem

The repo currently has 27 open PRs (as of 2026-07-01), almost all authored against asm/openapi.yaml, sat/openapi.yaml, or docs/index.html. A pass over the titles/diffs shows the same fixes were independently implemented multiple times, plus at least two PRs that are now stale because the same change already landed on main via a different PR. Merging any one PR in a cluster will put the others into conflict or make them redundant, and reviewer time is being spent evaluating N near-identical diffs instead of 1.

Duplicate / overlapping clusters

Theme	PRs	Notes
ASM Tags CRUD (GET/PATCH/DELETE /tags/{tag_id})	#3, #11, #22, #24	Four independent implementations of the same three endpoints. #22 and #24 are both full (GET+PATCH+DELETE) and based on the same commit (757d6e5) — they will conflict with each other.
ASM 403 Forbidden response coverage	#5, #9, #14, #20	Four PRs adding the same response to the same endpoints.
ASM 429/rate-limit documentation	#23 (headers only), #26 (429 responses), #31 (429 responses + headers)	#31 looks like a superset of #23 + #26.
OpenAPI 3.0 nullable: true → 3.1 type-union syntax	#4, #18	Same mechanical fix done twice.
Docs page Open Graph / Twitter / canonical meta tags	#13, #25, #32	Three independent implementations touching docs/index.html.
ASM response examples / descriptions	#7, #17, #28	Overlapping scope ("add response examples", "fix missing descriptions") — needs a diff comparison to confirm exact overlap.

Stale PRs (already superseded by main)

• feat: add SAT OpenAPI 3.1 spec #6 and feat: add SAT OpenAPI 3.1 specification #8 — both add sat/openapi.yaml from scratch. That file was already added to main in PR docs: sync SAT OpenAPI spec into api-docs and repair spec-lint CI #27 (e4f2088, merged 2026-06-30). Both PRs predate that merge and will now hard-conflict on the same new file.
• fix: pin Redoc CDN to @2 instead of @latest #12 — pins the Redoc CDN script to @2. main already contains "Pin Redoc and Spectral CLI to exact versions" (72a5ab9). Worth confirming fix: pin Redoc CDN to @2 instead of @latest #12 doesn't add anything beyond what's already merged before closing.

Suggested fix

1. For each cluster, pick the most complete/recent PR, merge it, and close the rest with a comment pointing to the survivor (Closing as duplicate of #NN).
2. Close feat: add SAT OpenAPI 3.1 spec #6 and feat: add SAT OpenAPI 3.1 specification #8 outright — superseded by the already-merged sat/openapi.yaml.
3. Verify and likely close fix: pin Redoc CDN to @2 instead of @latest #12 — appears already covered by 72a5ab9 on main.
4. To prevent recurrence: since these PRs are opened by an automated contributor (dmchaledev / Claude Code sessions), add a step (e.g. a CONTRIBUTING.md note or a pre-flight check in whatever triggers these runs) requiring a check of existing open PRs/issues touching the same file before starting new work.

Impact

Without triage, whichever PR in each cluster merges first will silently break the others (merge conflicts on asm/openapi.yaml/sat/openapi.yaml), and reviewers are currently facing ~3-4x the actual number of distinct changes to evaluate.

[dmchaledev]

Update (2026-07-06): problem has gotten worse, not better

Five days on, none of the clusters listed above have been triaged — every PR named in this issue (#3, #4, #5, #6, #7, #8, #9, #11, #12, #13, #14, #17, #18, #20, #22, #23, #24, #25, #26, #28) is still open. Meanwhile the queue grew from 27 to 30 open PRs, and the growth itself is now duplicating work:

• docs: add CONTRIBUTING.md and PR template to prevent duplicate PRs #35 ("add CONTRIBUTING.md and PR template to prevent duplicate PRs") already implements this issue's own suggested fix (item 4) — a pre-flight duplicate-check requirement in CONTRIBUTING.md + .github/pull_request_template.md. It's mergeable (mergeable_state: clean, docs-only, no CI required) and has been sitting unmerged since 2026-07-02. This should be the very first thing merged — every day it stays open, more duplicate PRs land that it would have prevented.
• security: add SRI hash to CDN-loaded Redoc script #19 and security: add SRI hash to Redoc CDN script tag #36 are now an additional duplicate pair: both add an SRI hash to the Redoc CDN <script> tag in docs/index.html.
• fix(sat): add missing 4xx/5xx error responses across the SAT API #37 implements the fix proposed in sat/openapi.yaml: add standard 4xx/5xx error responses to all endpoints #29 (SAT 4xx/5xx error responses) — not a duplicate, but another PR competing for review bandwidth while the backlog is unsorted.

Recommended next action, in order

1. Merge docs: add CONTRIBUTING.md and PR template to prevent duplicate PRs #35 now to stop the bleeding (adds the process guardrail; zero functional risk).
2. Work the existing clusters from this issue's table, picking one survivor each and closing the rest as duplicates: Tags CRUD (feat(asm): add PATCH and DELETE /tags/{tag_id} to complete the Tags resource #3, feat(asm): add GET and DELETE /tags/{tag_id} to complete Tags CRUD #11, feat(asm): complete Tags resource — add GET, PATCH, DELETE /tags/{tag_id} #22, feat(asm): add GET, PATCH, DELETE /tags/{tag_id} to complete Tags CRUD #24), 403 coverage (feat(asm): add 403 Forbidden response to all endpoints #5, feat(asm): add 403 Forbidden response to all ASM API endpoints #9, docs(asm): add 403 Forbidden response to all write operations #14, docs(asm): add 403 Forbidden response to all API endpoints #20), 429/rate-limit (docs(asm): document X-RateLimit-Limit/Remaining response headers on all endpoints #23, fix: add missing 429 responses to all ASM API endpoints #26, fix(asm): add missing 429 responses and document rate-limit headers #31), nullable 3.0→3.1 (fix(asm): replace OAS 3.0 nullable: true with OAS 3.1 type unions #4, fix: replace deprecated nullable:true with OpenAPI 3.1 null type syntax #18), OG/Twitter meta tags (docs: pin Redoc to v2 and add Open Graph / SEO meta tags #13, Add Open Graph and canonical meta tags to API docs page #25, docs: add Open Graph and Twitter card meta tags to docs page #32), response examples (docs(asm): add response body examples to all 11 operations #7, docs(asm): add response examples and fix missing operation descriptions #17, docs: add missing error examples in ASM spec and .gitignore #28).
3. Add the new SRI duplicate pair to the same triage pass: keep one of security: add SRI hash to CDN-loaded Redoc script #19 / security: add SRI hash to Redoc CDN script tag #36, close the other.
4. Close feat: add SAT OpenAPI 3.1 spec #6, feat: add SAT OpenAPI 3.1 specification #8 (superseded by sat/openapi.yaml already on main) and fix: pin Redoc CDN to @2 instead of @latest #12 (superseded by 72a5ab9) outright.

Filed no new issue/PR for this to avoid adding to the exact pile-up this issue describes — flagging here instead so the existing tracking issue reflects current state.

---

Generated by Claude Code

[dmchaledev]

Update (2026-07-18): still unmerged, pile-up has accelerated to 40 open PRs

Twelve days after the last update here, nothing in this issue has been actioned:

• docs: add CONTRIBUTING.md and PR template to prevent duplicate PRs #35 (the CONTRIBUTING.md / PR-template fix recommended as the Add ASM OpenAPI 3.1 spec and CI linting workflow #1 priority in the previous update) is still open, unmerged, still mergeable_state: clean, 16 days after it was opened.
• Because docs: add CONTRIBUTING.md and PR template to prevent duplicate PRs #35 never merged, the duplicate-PR pattern it was meant to stop has continued and gotten worse: there are now 40 open PRs (up from 27 on 2026-07-01, 30 on 2026-07-06).
• The SRI-hash fix for docs/index.html alone now has 7 duplicate PRs: security: add SRI hash to CDN-loaded Redoc script #19, security: add SRI hash to Redoc CDN script tag #36, security: add SRI hash to Redoc CDN script tag #43, Add SRI hash to the Redoc CDN script in docs/index.html #44, Add Subresource Integrity hash to Redoc CDN script #45, docs: add SRI hash to CDN-loaded Redoc script #46, docs: add SRI hash to Redoc CDN script tag #47, Add Subresource Integrity to the Redoc CDN script #48 (one new near-identical copy opened on almost every single day from 07-12 through 07-17).
• The SAT README quick-start fix now has 3 duplicates: docs: fix broken SAT quick-start example in README #40, docs: fix SAT quick-start example to match the real OpenAPI spec #41, docs: fix incorrect SAT quick-start example in README #42.
• All prior clusters from this issue's table (Tags CRUD, 403 coverage, 429/rate-limit, nullable 3.0→3.1, OG/Twitter meta tags, response examples) remain untriaged and still open.

The pattern of one new near-duplicate PR landing roughly every 24 hours strongly suggests this is coming from a recurring automated task (e.g. a scheduled "find one improvement and open a PR" run) that isn't checking the 30-40 already-open PRs before starting new work — the exact failure mode #35 was written to prevent.

Not opening a new PR/issue for this (that would just be pile-up #41) — flagging here again, and separately notifying the repo owner directly, since automated comments on this issue aren't getting picked up.

Recommended immediate actions, unchanged from before but now more urgent:

1. Merge docs: add CONTRIBUTING.md and PR template to prevent duplicate PRs #35 — it's the guardrail that stops new duplicates from landing.
2. Whatever schedule/automation is opening one PR/day against this repo should be paused until the existing backlog is triaged.
3. Triage and close the duplicate clusters (list above and in the original issue body).

---

Generated by Claude Code

[dmchaledev]

Resolved: triaged all 40 open PRs (28 duplicate-cluster members closed, 12 distinct fixes merged after resolving conflicts as needed) — the repo is back to 0 open PRs. CONTRIBUTING.md + PR template from #35 (merged) are now in place to prevent recurrence.

---

Generated by Claude Code