Skip to content

Skill idea: triage-build — classify a failed lecture build against known failure patterns #37

Description

@mmcky

Parked skill idea, moved from FUTURE-IDEAS.md when that file was retired in favour of tracker issues (2026-08-13 maintenance round). Evidence base: the 2026-07 PR-history analysis (~630 merged PRs across the four main lecture repos).

The idea

Classify a failed lecture build against known failure patterns — deprecation sweeps, pandas 3.0 read-only arrays, unpinned-package breakage à la prettytable, SymPy overflow — and propose the known fix. Recurs roughly 15×/year across repos.

Assessment (2026-08-13)

Real and recurring, but it overlaps the jupyter-book-fixer idea in QuantEcon/meta#304, so the first step is coordination, not code: decide whether this lives here as a skill or in the CI tooling that meta#304 sketches. A skill has one genuine advantage — the pattern library (failure signature → known fix) is institutional knowledge of exactly the kind this marketplace exists to share, and it can run interactively on a build log an author pastes in, not only in CI.

The pattern library is also the maintenance cost: it goes stale with every environment upgrade cycle, so the skill is only worth building with a named maintainer for that list.

Metadata

Metadata

Assignees

No one assigned

    Labels

    enhancementImprovement to existing content or functionalitylow-priorityNice to have, no time pressure

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions