docker: production image + Gitea Actions for monthly refresh

Dockerfile: self-contained image with corpus + Chroma + BM25 baked
in. Drawbar's compose pulls + runs without volume mounts. Built from
sources.json (labels schema), PRODUCT_NAME=crop_chem by default,
HYBRID_SEARCH=true (always-on for production quality). RERANK_URL +
OLLAMA_URL get set at compose time.

.gitea/workflows/refresh.yml: monthly cron (1st @ 06:00 UTC) does
full scrape → reindex → image push. Scrapes Bayer (~30 min) +
EPA PPLS row-crop filtered (~7h). Skips reindex+push if no corpus
diff. Tags pushed: :latest, :<sha12>, :corpus-<YYYY.MM.DD>.

.gitea/workflows/image-only.yml: on-demand or auto on code-only
pushes to main (paths: docs_mcp/, rag/, scrape/, requirements.txt,
Dockerfile, sources.json). Reindexes from committed corpus, builds
image, pushes. ~10 min vs ~9h full refresh.

.gitignore: corpus/ now COMMITTED (4,159 labels, 265 MB of .md +
sidecars). Lets image-only.yml rebuild indexes without re-scraping.
chroma/ + bm25/ still gitignored (regenerable binary indexes).

.dockerignore: drops venv, eval results, PLAN/README/CLAUDE.md,
deploy/, .git/ — keeps the image lean. corpus + chroma + bm25
explicitly NOT in dockerignore (those go INTO the image).

Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>

corpus/epa_ppls/5481-524.md

@@ -0,0 +1,152 @@
+# IMPACT HERBICIDE
+
+- EPA Reg No: **5481-524**
+- Registrant: AMVAC CHEMICAL CORPORATION
+- Signal word: Caution
+- Active ingredients: Topramezone (29.7%)
+- Label accepted: 2021-03-11
+- Source PDF: https://www3.epa.gov/pesticides/chem_search/ppls/005481-00524-20210311.pdf
+
+---
+524-20210310 Page 1 of 3 
+IMPACT Herbicide
+EPA REG. NO. 5481-524 
+ 
+SUPPLEMENTAL LABEL 
+For control of emerged weeds 
+in all corn types
+ 
+SUPPLEMENTAL LABEL 
+Increased maximum single and annual use rate in corn to 2.0 fluid ounce 
+(0.044 lb topramezone) per acre. 
+EPA Reg. No.: 5481-524
+Directions for Use 
+This supplemental label expires April 1, 2024 and must not be used or distributed after this 
+date. 
+ 
+It is a violation of Federal law to use this product in a manner inconsistent with its labeling. 
+ 
+THIS LABEL MUST BE IN THE POSSESSION OF THE USER AT THE TIME OF APPLICATION. 
+ 
+Before applying IMPACT please read the complete label affixed to the product container. All 
+applicable directions, restrictions and precautions on the EPA-registered label are to be followed. Use 
+of IMPACT according to this supplemental labeling is subject to the use precautions and limitations 
+imposed by the label on the container. 
+ 
+IMPORTANT: Read the entire Directions for Use and the Limited Warranty and Disclaimer on the EPA-
+registered IMPACT herbicide label before using this product. 
+ 
+Use Information 
+For use in FIELD CORN (grown for grain, silage or seed), POPCORN (ear, kernel or seed) and SWEET 
+CORN (ear, kernel or seed). 
+Table 1. Emerged Broadleaf Weeds Controlled or Partially Controlled with IMPACT and Maximum 
+Weed Size at Application 
+Broadleaf Weeds1
+IMPACT 
+0.75 fl oz per acre 
+IMPACT 
+1.0 fl oz per acre 
+2 2 
+Amaranth, Palmer C C 
+Amaranth, Powell C C
+Bindweed, field PC PC
+
+524-20210310 Page 2 of 3 
+Broadleaf Weeds1
+IMPACT
+0.75 fl oz per acre
+IMPACT
+1.0 fl oz per acre
+2 2 
+Burcucumber C 3,4 C 4 
+Buckwheat, wild NC C 3,4
+Canola, volunteer C C
+Carpetweed C C
+Chickweed, common C C 
+Cocklebur, common C C 
+Dandelion PC PC
+Devil’s-claw C 3,4 C 4 
+Galinsoga, hairy C C 
+Henbit C C 
+Jimsonweed C C 
+Kochia C 4 C 4 
+Ladysthumb C 4 C 4 
+Lambsquarters, common C C 
+Lettuce, prickly C C
+Mallow, common C 3,4 C3,4
+Mallow, Venice C 3,4 C 3,4
+Marestail (horseweed) C C
+Morningglory, entireleaf C 3,4 C 3,4
+Morningglory, ivyleaf C 3,4 C 3,4
+Mustard, wild C C 
+Nightshade, black C C
+Nightshade, Eastern black C C
+Nightshade, hairy C C
+Pigweed, prostrate C C
+Pigweed, redroot C C
+Pigweed, smooth C C
+Pigweed, tumble C C
+Purslane, common PC 3 C3,4
+Pusley, Florida C 3 C
+Ragweed, common C C
+Ragweed, giant C C
+Shepherd’s-purse C C
+Sida, prickly C 3,4 C 4 
+Smartweed, Pennsylvania C 4 C 4 
+Sunflower, wild (common) C C
+Thistle, Canada PC 3 PC 
+Thistle, Russian C C
+Velvetleaf C C
+Waterhemp, common C C
+Waterhemp, tall C C
+1Refer to Section X for a list of scientific names on the EPA label. 
+2Broadleaf Weed Rating: C=Control; PC=Partial Control; NC=Not Controlled. 
+3Apply before weed exceeds 3 inches in height. 
+4 rol.
+
+524-20210310 Page 3 of 3 
+Table 2. Emerged Grass Weeds Controlled or Partially Controlled with IMPACT at 0.75 to 2.0 fl oz/A 
+and Maximum Weed Size at Application 
+Grass Weeds1
+IMPACT 
+0.75 fl oz per acre 
+IMPACT 
+1.0 fl oz per acre 
+IMPACT 
+1.5 – 2.0 fl oz per acre 
+2 2 2
+Barnyardgrass C 5 C C 
+Crabgrass, large C 5 C 5 C 
+Crabgrass, smooth C 5 C 5 C 
+Cupgrass, woolly PC C 3 C 
+Foxtail, giant C C C 
+Foxtail, green C 5 C C 
+Foxtail, yellow PC C 3,5 C 
+Goosegrass C C C 
+NC NC PC 4 
+Johnsongrass, seedling PC C C
+Millet, wild-proso C C C 
+Panicum, fall PC C 3,5 C 
+Panicum, Texas PC PC C 4,5 
+Sandbur, field NC PC C 4,5 
+Shattercane PC PC C 4,5 
+Signalgrass, broadleaf PC C 3,5 C 5
+1Refer to Section X for a list of scientific names on the EPA label. 
+2Grass Weed Rating: C=Control; PC=Partial Control; NC = Not Controlled. 
+3Apply before grass exceeds 3 inches in height. 
+4 
+5 
+ 
+Restrictions 
+ the 
+equivalent of 2.0 fluid ounces of IMPACT per acre. 
+ DO NOT apply IMPACT within 45 days of corn harvest (fresh market sweet corn, silage, fodder, 
+or grain). 
+ an 
+application of IMPACT. 
+ DO NOT apply this product through any type of irrigation system. 
+©2020 AMVAC Chemical Corporation is a wholly owned subsidiary of American Vanguard Corporation. All Rights Reserved. 
+AMVAC, Impact, and their respective logos are trademarks owned by AMVAC Chemical Corporation. 
+Registrant: 
+AMVAC Chemical Corporation 
+MacArthur Court, Suite 1200