// Use this skill to update Super Sirius documentation after code changes. Trigger when the user says "update docs", "refresh documentation", "sync docs with code changes", or after merging PRs that changed the Super Sirius codebase. Inspects merged PRs since the last update and patches affected doc files.
| name | update-docs |
| description | Use this skill to update Super Sirius documentation after code changes. Trigger when the user says "update docs", "refresh documentation", "sync docs with code changes", or after merging PRs that changed the Super Sirius codebase. Inspects merged PRs since the last update and patches affected doc files. |
Read docs/super-sirius/README.md and extract the commit hash from the HTML comment at the bottom:
<!-- last-updated-commit: COMMIT_HASH -->
If no marker exists, ask the user which commit to start from, or default to the initial docs commit.
Run git log --oneline --merges LAST_COMMIT..HEAD to find merge commits.
Run gh pr list --state merged --limit 100 --base dev to list merged PRs.
Cross-reference to find PRs merged since the last update.
For each PR found:
gh pr view PR_NUMBER to get title, description, and labelsgh pr diff PR_NUMBER --name-only to get changed filesreferences/change-classification-guide.mdFor each classified change:
operators.md; if custom get_next_task_hint(), also update task-creator.mdoptimizations.md with: PR#, motivation, mechanism, code path, configphysical-plan-generation.md Part 3configuration.mdCRITICAL: Read the actual source code for any changed files before updating documentation. Do not guess based on PR titles alone.
IMPORTANT: Present proposed changes one file at a time for user review. Do not apply all changes at once. Show the user what will change in each file and wait for approval before proceeding to the next file.
STYLE: Do not use changelog language ("replaced X with Y", "extracted from", "previously X, now Y"). Describe the current design as-is. These are reference docs, not a changelog. Only include PR numbers in optimizations.md entries.
DEPTH: Keep descriptions at the right level of abstraction for the doc file. Don't add overly detailed implementation specifics (e.g., specific deadlock fixes, internal bug fixes) — focus on concepts, interfaces, and behavior that a developer reading the docs needs to understand.
Replace the commit hash in docs/super-sirius/README.md:
<!-- last-updated-commit: NEW_HEAD_HASH -->
Print a summary of:
Example 1: User says "update the docs"
abc1234src/op/sirius_physical_hash_join.cpp -> updates operators.mdoptimizations.mdExample 2: User says "update docs" but no changes affect Super Sirius
src/ modificationsIf docs/super-sirius/README.md doesn't contain the marker comment, ask the user which commit to use as baseline, or use git log --oneline docs/super-sirius/ | head -1 to find when docs were last modified.
For PRs with 50+ changed files, focus only on files matching the source paths in the classification guide. Skip test files, build files, and third-party changes.
If a PR touches multiple areas, update all affected docs. If unsure whether a change warrants a doc update, err on the side of updating.
Use this skill when the build fails, compilation errors occur, or you see undefined references, linker errors, CUDA compilation issues, missing headers, or template instantiation failures. Analyzes errors, suggests fixes, and iteratively rebuilds until success.
Run TPC-H benchmarks on Super Sirius or DuckDB CPU baseline — generate data, execute queries, validate results, and compare timings. Trigger when the user mentions benchmarking, TPC-H, performance testing, query runtimes, or wants to compare Sirius vs DuckDB speed.
Discover and document a dependency library or submodule — analyzes all uses within the codebase, divides the library into logical modules, identifies which modules are used, and generates LLM-consumable API documentation for each module. Use when the user wants to understand a library dependency, map its modules, or generate API reference docs for a submodule.
Use this skill when a Sirius query crashes, segfaults, hangs, throws an exception, or unexpectedly falls back to CPU. Also use when you see CUDA errors, std::bad_alloc, or the process gets killed. Diagnoses issues using log analysis, cuda-gdb, AddressSanitizer, and NVIDIA Compute Sanitizer.
Use this skill when a Sirius query returns wrong results, missing rows, extra rows, or incorrect values compared to DuckDB CPU. Pinpoints the faulty operator using per-operator row counts and data checksums. Also detects CUDA stream synchronization issues that cause garbage data.
Use this skill to generate benchmark datasets (TPC-H, TPC-DS, etc.). Trigger when the user needs test data at a specific scale factor for benchmarking or testing. Supports parquet and duckdb output formats.