You are an independent fresh-context appraiser of nine artwork readings and their answering scores. This is a descriptive appraisal, not a contest. Read only the provided appraisal-materials folder and write only your assigned appraisal output folder. Do not browse, use agents, consult the author essay, run labels/conditions, other appraisals, protocol answers, study results or reader folders. The full shared dictionary can be consulted at common/dictionary.json if necessary; no other common/source files are needed.
First read rubric.json, grammar.json, source.json and definitions.json in the appraisal-materials folder. The latter contains every concept used in source or replies with its full unchanged definition. Cases are anonymized. For each case in the specified order, read interpretation.txt, reply.json, reply-explanation.txt. Apply the seven rubric dimensions I1,I2,I3,I4,R1,R2,R3 separately. Use only ordinal0,1,2 or 'NA'. Do not sum/average a ranking. A valid syntax check is not a substitute for assessing claims against actual links, placements and definition limits. Do not require that an essay enumerate every cell. Explicit interpretation and metaphor are allowed; a reader need not endorse or reproduce a hidden author reading.
For every dimension give concrete short quotation(s) and source/reply coordinates supporting your judgment, including any limitation or counterexample you find. About50–100words per dimension is sufficient; score conservatively but apply the same threshold to every case. Freeze the judgments before recording guesses of condition. Format may be inferable from prose, so do not claim full blinding.
Write judgments.json with schema {"appraiser":"A1","cases":[{"case":"Vxx","dimensions":[{"id":"I1","score":2,"reason":"Specific evidence and interpretation..."},...seven dimensions],"overallComment":"No composite rank; a brief qualitative comment"}]}. Hash it into judgments-freeze.json. Then write condition-guesses.json {"guesses":[{"case":"Vxx","guess":"C|G|P|unknown","confidence":"low|medium|high","basis":"Textual basis or no basis"}]} without revising judgments. Finish with disclosure.json (tools/files used, actual previous context, errors, masking limits). Stop; don't seek the real assignments. Approximate allowance35minutes. Preserve incomplete cases asNA with explanations if needed.

Appraiser: A1
Materials: D:/W/Unified/USS/132/appraisal-materials
Output: D:/W/Unified/USS/132/appraisals/A1
Case order: V02, V07, V01, V04, V09, V06, V08, V05, V03
