Skip to content

ericrisco/rsc-harness / skills/duckdb

duckdb

Flagged

Use when analytical SQL must run in-process with no server: Parquet/CSV/JSON/Arrow queried in place, OLAP embedded in an app or notebook, a slow pandas groupby on multi-GB data, or S3/lakehouse data read without downloading. NOT a multi-user analytics server (that is clickhouse-analytics), NOT an app's transactional CRUD store (that is postgresdb).

FileKindSize
SKILL.mdskill md10.5 KB
evals/README.mdmarkdown2.0 KB
evals/cases.yamlconfig4.0 KB
references/python-and-interop.mdmarkdown4.0 KB
references/remote-and-lakehouse.mdmarkdown3.1 KB
scripts/verify.shexecutablescript2.6 KB

evals/README.md

View on GitHub
# Eval harness — `duckdb`

This checks two things: **triggering** (does the skill fire on real DuckDB situations and stay quiet on
near-misses that belong to a sibling?) and **capability** (does loading the skill measurably improve the
answer?). Cases live in `cases.yaml`. These are graded by a human or a driver agent feeding prompts to a
fresh session — not an automated runner. The executable check is `scripts/verify.sh`, which smoke-tests
that the documented DuckDB commands actually run; the eval cases here grade judgment, not code.

## Triggering

For each prompt: start a fresh agent with only the `duckdb` skill discoverable, paste the prompt verbatim,
and observe whether the agent reaches for DuckDB. Run 3–5 trials (the decision is stochastic).
`should_trigger` passes if DuckDB fires in the majority of trials. `should_not_trigger` passes if it does
**not** fire; each such case names a `route_to` sibling (`clickhouse-analytics`, `postgresdb`, `sql`,
`sqlite-turso`, `vector-db`) — sanity-check the agent would plausibly route there, but any not-duckdb route
still passes this skill's gate. Some routed siblings may not be built in this collection yet; the routing
intent is what's graded. Pass bar: ≥ 90% trigger accuracy across all cases.

## Capability

Run the scenario twice — WITHOUT the skill (clean agent) and WITH it — and grade each transcript against
the `must_include` checklist, one point per bullet that is specifically and correctly covered (not just
name-dropped). WITH the skill should cover ≥ 80% of bullets and clearly beat WITHOUT (target ≥ +30 points
or crossing fail→pass). If a baseline agent already nails it, tighten the rubric.

## Notes

Prompts are deliberately varied: some never say "DuckDB" (slow pandas groupby, out-of-core join,
analyze-a-CSV-without-a-server), and one is in Spanish — these test intent recognition, not keyword
matching. Record trial counts and per-bullet verdicts so a reviewer can audit; don't report a bare
pass/fail.
deprecation-and-migrationaddyosmani/agent-skillsManages deprecation and migration. Use when removing old systems, APIs, or features. Use when migrating users from one implementation to another. Use when migrating a database schema in production, such as renaming or dropping a column without downtime (expand/contract). Use when deciding whether to maintain or sunset existing code.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
performance-optimizationaddyosmani/agent-skillsOptimizes application performance across frontend, backend, queries, and databases. Use when performance requirements exist, when you suspect performance regressions, when Core Web Vitals or load times need improvement, when N+1 query patterns need fixing, or when profiling reveals bottlenecks.Stars: 119kUpdated Claude CodeCodexVerifiedInstall
agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json.Stars: 118kUpdated Claude CodeCodexVerifiedInstall
agent-cost-reportthedotmack/claude-memBelievable agent cost report for any period, default the last 7 full days PT, not counting today. Measured tokens from Claude Code transcripts priced at OpenRouter list prices (ESTIMATED), measured provider spend when a sanctioned source exists, note-taker cost separate, Timing-style HTML/PDF plus report.json, line-items.csv, evidence.json.Stars: 118kUpdated Claude CodeCodexVerifiedInstall