pg-shadow-implementlisted

Shadow-implement a pgsql-hackers thread to calibrate the pg-claude planner suite — read the COVER + discussion only (NOT the attached patch), produce a spec, run pg-feature-brainstorm / pg-feature-plan / pg-implement as-if the user asked for the feature, then fetch the upstream patch, diff against ours, score the gap, and emit comparison.md + skill-gaps.md. This is the "Phase E" calibration loop. Use when the user invokes `/pg-shadow <thread-url>`, says "shadow-implement that thread", "run Phase E against <thread>", "run a shadow against this hackers thread", or names a pgsql-hackers thread + asks how our planner would handle it. Do NOT trigger for real upstream patches we plan to send (use `pg-feature-plan` + `pg-implement` directly), for already-committed threads (re-implementing landed code is meaningless), or for non-PG calibration.
matejformanek/postgres-claude · ★ 0 · AI & Automation · score 70

Install: claude install-skill matejformanek/postgres-claude

# pg-shadow-implement — Phase E shadow-implementation runner This skill operationalizes `knowledge/calibration/shadow-implementation-methodology.md` as a runnable procedure. Each run produces calibration data about the planner suite's implementation accuracy, NOT a patch we plan to send upstream. The pairing: - **Phase C** calibrated review (`pg-patch-review` / `review-checklist`). - **Phase E** (this skill) calibrates implementation (`pg-feature-plan` / `pg-implement`). ## When to use vs not | Situation | Use | |---|---| | Active pgsql-hackers thread with COVER + attached patch + clear scope | **this skill** | | Already-committed patch (we'd be re-implementing) | neither — useless | | Thread we plan to send patches against | `pg-feature-plan` directly (no shadow) | | Multi-thread series where the COVER doesn't fully specify scope | neither — too much guessing | | Touched subsystem isn't documented in our corpus | neither — guessing not calibrating | ## Inputs - **Thread URL** (required) — a `https://www.postgresql.org/message-id/...` link to the thread's COVER message, or the `flat` view URL. - **Slug** (optional) — short kebab-case name for the run directory (auto-derived from the thread subject if omitted). ## Output Per run, under `knowledge/shadow-implementations/<slug>/`: - `spec.md` (~50-200 LOC) — COVER extraction, context probe, engagement classification. - `plan.md` (~200 LOC) — produced by `pg-feature-plan`. - `notes.md` (~100 LOC) — produced by `p