From document-processing
Update an existing processed document in 3-output/ - add a new source to 1-input/, apply corrections found in the output, re-apply changed uniformization rules, or re-verify against updated sources. Always re-runs the grounding CLI on the changed content before declaring done. Use when asked to update the document, add a new source to the timeline, re-verify after sources changed, or apply corrections to an output.
How this skill is triggered — by the user, by Claude, or both
Slash command
/document-processing:updateThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Update an existing processed document with new information, corrections, or re-verification against updated sources. **Every update ends with a mandatory CLI-grounding pass on the changed content** - no silent skip - because any content change can introduce an ungrounded claim.
Update an existing processed document with new information, corrections, or re-verification against updated sources. Every update ends with a mandatory CLI-grounding pass on the changed content - no silent skip - because any content change can introduce an ungrounded claim.
python3 -c "import stellars_claude_code_plugins" 2>/dev/null || python3 -m pip install --user --upgrade stellars-claude-code-plugins
No-op when the package is importable; auto-installs when missing OR when a stale shim is on PATH but the package is uninstalled in the active Python. Never ask - just run the line.
1-input/ - re-run affected sections3-output/ document - correct and re-verifyIdentify what exists: Read 3-output/ for the current document, 2-wip/ for the processing history, INSTRUCTIONS.md and BENCHMARK.md for the original program.
ASK user:
Targeted update (default - faster):
2-wip/<task-name>/<filename>_prev.md first)Full re-run (if structural changes needed):
INSTRUCTIONS.md with new contextprocess skill using existing + new sources3-output/MANDATORY grounding pass (every update, no exception): run the grounding CLI on the updated content via the grounding skill -
document-processing extract-claims --document <updated doc> --output validation/claims.json, review validation/claims.jsondocument-processing batch-ground --claims validation/claims.json --source <src1> --source <src2> ... --output validation/grounding-report.md (one --source per 1-input/ source; pass --semantic on — the default — unless the user has opted out, i.e. settings.semantic_enabled == false; see the grounding skill)document-processing check-consistency --document <updated doc> --output validation/consistency-report.md
Fix any UNCONFIRMED / CONTRADICTED claim and any consistency finding before declaring done. Apply the grounding skill's verdict rules.Re-verify against benchmark: Run BENCHMARK.md evaluation against the updated document. Report the score delta.
2-wip/<task-name>/<filename>_prev.md1-input/ first - never process from arbitrary locations1-input/ stays read-only; never modify source materialgrounding skill - this skill calls them, never re-implements themnpx claudepluginhub stellarshenson/claude-code-plugins --plugin document-processingProvides a checklist for code reviews covering functionality, security, performance, maintainability, tests, and quality. Use for pull requests, audits, team standards, and developer training.