| name | sync-corpus |
| description | Run the full corpus sync pipeline in safe sequence. Use when the user says "sync", "sincronizar", "pipeline sync", "atualizar corpus", or after adding new candidates to corpus-data.json.
|
| user-invocable | true |
Run the ICONOCRACIA corpus sync pipeline in sequence. Stop immediately if any step fails.
Pipeline
| Step | Script | What it does |
|---|
| 1 | validate_schemas.py | Validates records.jsonl against master-record schema |
| 2 | vault_sync.py status | Reports vault ↔ records.jsonl drift |
| 3 | records_to_corpus.py --diff | Reports records.jsonl ↔ corpus-data.json drift |
| 4 | sync_companion.py | Rebuilds companion-data.json from corpus |
| 5 | refresh_dashboard.py | Re-embeds data into HTML dashboards |
| 6 | code_purification.py --status | Reports endurecimento scores across corpus |
Execution
set -e
REPO="$CLAUDE_PROJECT_DIR"
PYTHON="conda run -n iconocracy python"
echo "=== Sync Corpus ICONOCRACIA ==="
echo "[1/6] Validando schemas..."
$PYTHON "$REPO/tools/scripts/validate_schemas.py"
echo "[2/6] Vault drift..."
$PYTHON "$REPO/tools/scripts/vault_sync.py" status
echo "[3/6] Records ↔ corpus drift..."
$PYTHON "$REPO/tools/scripts/records_to_corpus.py" --diff
echo "[4/6] Sincronizando companion data..."
$PYTHON "$REPO/tools/scripts/sync_companion.py" > /dev/null
echo "[5/6] Atualizando dashboards..."
$PYTHON "$REPO/tools/scripts/refresh_dashboard.py"
echo "[6/6] Status de purificacao..."
$PYTHON "$REPO/tools/scripts/code_purification.py" --status
Report format
After running, present a table:
| Passo | Status | Detalhes |
|---|
| validate_schemas | OK/FAIL | N itens válidos / erro |
| vault_sync | OK/DRIFT | N notas, X SCOUT pendentes |
| records_to_corpus | OK/DRIFT | N itens só em records / M só em corpus |
| sync_companion | OK/FAIL | companion-data.json atualizado |
| refresh_dashboard | OK/FAIL | corpus + agents dashboards |
| code_purification | OK/FAIL | Score médio endurecimento: X.X |
If any step fails, show the full error and stop. Do not run subsequent steps.
Log the run via tools/scripts/log_agent_run.py --agent sync --status success/error.
When to run
- After adding new items to
corpus/corpus-data.json
- After running
ingest.py on a new batch of scans
- Before committing changes to the corpus
- After running Scout campaigns that modify corpus data
Notes
- Bridge step (
corpus_bridge.py) removed — module deleted, ingest pipeline
refactor in progress (2026-04).
make_index.py is the Iconclass index builder (takes <lang> <cmd>),
not a corpus tool — excluded.
