Agent tooling fixes: URL scrape, organize docs, summary urls (0.1.17–0.1.19) #1

Merged

ValleyGeek merged 3 commits from 0.1.16/mcp-agent-tooling-fixes into main

2026-08-06 03:15:49 +00:00

Grok commented

2026-08-06 03:15:30 +00:00

Member

Summary

Three scoped MCP façade improvements from the agent organize-workflow review — only changes judged in-scope for sa-mcp (no host-side analytics or external scraping).

Version	Change
0.1.17	Fix `scraper_run` `kind=url` with typed `scrape*URL` field selections; default `content_type=SCENE`
0.1.18	Docs: organize/gap filters, path vs regex, `urls`/`tag_ids` replace vs bulk ADD/SET/REMOVE
0.1.19	Scene `detail=summary` includes `urls` (list finds + write ACKs)

Details

0.1.17 — URL scrape fix

Bare scrapeURL without subfields was invalid GraphQL (union).
Routes by content_type to scrapeSceneURL / scrapePerformerURL / gallery / image / group with explicit selections.
Validates non-empty url; rejects unknown content types; MOVIE → GROUP.

0.1.18 — Agent docs

Common filters for unorganized / missing performers / missing url / path.
Path INCLUDES vs scene_find_by_path regex guidance.
Documents that scene_update / scene_update_many replace urls and tag_ids; bulk supports ADD/REMOVE/SET.

0.1.19 — Summary ACKs

SCENE_SUMMARY adds urls so agents can verify URL writes without a follow-up scene_get.

Test plan

uv run pytest tests/test_scraper_run_url.py tests/test_scene_detail.py tests/test_scene_update_many.py -q
uv run ruff check on touched paths
Optional: live Stash scraper_run kind=url against a known content URL if scrapers installed

## Summary Three scoped MCP façade improvements from the agent organize-workflow review — only changes judged in-scope for sa-mcp (no host-side analytics or external scraping). | Version | Change | |---------|--------| | **0.1.17** | Fix `scraper_run` `kind=url` with typed `scrape*URL` field selections; default `content_type=SCENE` | | **0.1.18** | Docs: organize/gap filters, path vs regex, `urls`/`tag_ids` replace vs bulk ADD/SET/REMOVE | | **0.1.19** | Scene `detail=summary` includes `urls` (list finds + write ACKs) | ## Details ### 0.1.17 — URL scrape fix - Bare `scrapeURL` without subfields was invalid GraphQL (union). - Routes by `content_type` to `scrapeSceneURL` / `scrapePerformerURL` / gallery / image / group with explicit selections. - Validates non-empty `url`; rejects unknown content types; `MOVIE` → `GROUP`. ### 0.1.18 — Agent docs - Common filters for unorganized / missing performers / missing url / path. - Path `INCLUDES` vs `scene_find_by_path` regex guidance. - Documents that `scene_update` / `scene_update_many` **replace** `urls` and `tag_ids`; bulk supports ADD/REMOVE/SET. ### 0.1.19 — Summary ACKs - `SCENE_SUMMARY` adds `urls` so agents can verify URL writes without a follow-up `scene_get`. ## Test plan - [x] `uv run pytest tests/test_scraper_run_url.py tests/test_scene_detail.py tests/test_scene_update_many.py -q` - [x] `uv run ruff check` on touched paths - [ ] Optional: live Stash `scraper_run` kind=url against a known content URL if scrapers installed

Grok added 3 commits

2026-08-06 03:15:30 +00:00

fix(scraper): typed URL scrape selection sets (0.1.17) d9ebfeebb9

scraper_run kind=url used bare scrapeURL without subfields, which Stash
rejects. Route by content_type to scrapeSceneURL (default SCENE) and peer
typed fields with explicit selections; validate url and content_type.

docs: scene organize filters and write field semantics (0.1.18) 63f77ec108

Document common gap-fill filters, path INCLUDES vs basename regex, and
that scene_update/update_many replace urls and tag_ids while bulk supports
ADD/REMOVE/SET. Point architecture batch notes at tools.md.

feat(scenes): include urls in summary detail (0.1.19) 713636303a

SCENE_SUMMARY (list finds and scene update/create write ACKs) now returns
urls alongside tags/performers so agents can verify URL metadata patches
without a follow-up scene_get.