All settings can be overridden with DOKIMOS_-prefixed environment variables.
| Variable | Default | Description |
|---|
DOKIMOS_LOG_LEVEL | INFO | Default application log level |
DOKIMOS_CORPUS_PATH | corpus | Directory used for the internal source corpus |
DOKIMOS_INDEX_FILE | corpus/index.json | JSON index file path |
DOKIMOS_OUTPUT_DIR | output | Default output directory for reports |
| Variable | Default | Description |
|---|
DOKIMOS_CHUNK_STRATEGY | paragraph | One of paragraph, sentence, fixed |
DOKIMOS_CHUNK_SIZE | 500 | Approximate words per chunk for fixed chunking |
DOKIMOS_CHUNK_OVERLAP | 50 | Overlap between fixed chunks |
| Variable | Default | Description |
|---|
DOKIMOS_PLAGIARISM_BACKEND | hybrid | One of local, remote, or hybrid |
DOKIMOS_SHINGLE_SIZE | 5 | Number of words per shingle |
DOKIMOS_PLAGIARISM_JACCARD_THRESHOLD | 0.10 | Minimum Jaccard similarity to retain a candidate |
DOKIMOS_PLAGIARISM_FUZZ_THRESHOLD | 60.0 | Minimum RapidFuzz score to keep a reranked match |
DOKIMOS_PLAGIARISM_REMOTE_QUERY_MAX_CHUNKS | 5 | Maximum input chunks used as remote search queries |
DOKIMOS_PLAGIARISM_REMOTE_QUERY_MAX_CHARS | 180 | Maximum characters sent in each remote query |
DOKIMOS_PLAGIARISM_REMOTE_PER_PROVIDER_RESULTS | 3 | Maximum candidate sources requested from each provider per query |
DOKIMOS_PLAGIARISM_REMOTE_TIMEOUT_SECONDS | 10.0 | Remote request timeout |
DOKIMOS_PLAGIARISM_REMOTE_MAX_SOURCE_CHARS | 20000 | Maximum retained text size per fetched remote source |
DOKIMOS_PLAGIARISM_REMOTE_FETCH_FULL_TEXT | true | Attempt to fetch source pages or PDFs instead of relying on metadata |
DOKIMOS_PLAGIARISM_REMOTE_CONTACT_EMAIL | unset | Optional contact email included where provider etiquette recommends it |
DOKIMOS_PLAGIARISM_REMOTE_ENABLE_OPENALEX | true | Enable OpenAlex provider |
DOKIMOS_PLAGIARISM_REMOTE_ENABLE_CROSSREF | true | Enable Crossref provider |
DOKIMOS_PLAGIARISM_REMOTE_ENABLE_ARXIV | true | Enable arXiv provider |
DOKIMOS_PLAGIARISM_REMOTE_ENABLE_DUCKDUCKGO | true | Enable DuckDuckGo HTML web-search provider |
| Variable | Default | Description |
|---|
DOKIMOS_AI_SHORT_SENTENCE_THRESHOLD | 8 | Sentences at or below this word count are considered short |
DOKIMOS_AI_LONG_SENTENCE_THRESHOLD | 40 | Sentences at or above this word count are considered long |
DOKIMOS_AI_SIGNAL_TRIGGER_THRESHOLD | 0.5 | Per-signal threshold to mark a signal as triggered |
DOKIMOS_AI_RISK_HIGH_THRESHOLD | 0.6 | Aggregate score threshold for high risk |
DOKIMOS_AI_RISK_MEDIUM_THRESHOLD | 0.3 | Aggregate score threshold for medium risk |
DOKIMOS_AI_SHORT_DOCUMENT_WORDS | 80 | Documents below this size receive the short-document caveat |
dokimos analyze submissions/paper-01.pdf --format json
export DOKIMOS_INDEX_FILE=/tmp/course-a-index.json
dokimos index-sources ./course-a-corpus
dokimos analyze submissions/paper-01.pdf --format json
export DOKIMOS_PLAGIARISM_BACKEND=remote
dokimos analyze essay.txt --format json
export DOKIMOS_PLAGIARISM_BACKEND=local
dokimos index-sources ./course-a-corpus
dokimos analyze essay.txt --format json
export DOKIMOS_CHUNK_STRATEGY=sentence
dokimos analyze essay.txt --format json
dokimos analyze essay.txt --json-out reports/essay.json