Skip to content
SkillModel-led414 tokens in use

prepare-regression

The preparation stage of regression-hunt: pin before and after sources for a PR or commit and freeze its messages and discussion.

8,319 stars on the repository CC-BY-SA-4.0trailofbits/skills · plugins/regression-hunt/skills/prepare-regression
/plugin marketplace add trailofbits/skills
View source

Install command as written in the repository's README

Skill label

measured from its files

What installing prepare-regression costs

23tokens idle
414tokens in use

Idle is what sits in context while the skill is installed. In use is its instructions, loaded when it runs. Counts are approximate, made offline with cl100k_base.

Model dependence Model-led

Script-ledMixedModel-led

Mostly instructions, so how well it works depends on the model that follows them.

This describes the skill's structure, how much of it is scripts. It does not say how well it works.

Touches

shelldetected

Declared tags come from the skill's frontmatter; detected ones from its scripts, its code blocks and the commands it names in inline code. This is detected, not audited.

How it scores

Quality

How well the skill is written, counted from its own files: its frontmatter, its steps, examples, stated output and constraints, its size and whether every file it ships is one it names.How we score it →
Fair
  • Frontmatter12/15
  • Structure17/25
  • Examples3/15
  • Output7/15
  • Size10/15
  • Resources10/15

Security

Starts at 100 and loses points for each thing in the skill's files that a reader should look at first.How we score it →
Excellent

No findings.

Repository health

Pre-computed score based on development activity, maintenance, community, maturity, and trend momentum.How we score it →
Good
  • Development activity100
  • Maintenance52
  • Community64
  • Maturity32
  • Momentum40

Prepare-regression takes a PR URL, or a repository plus a commit or range, and works out the complete change as two states, S0 before and S1 after. It handles unmerged pull requests, merge commits, squash merges and rebased series, and blocks when a range is ambiguous or a root baseline is missing.

The agent freezes commit messages or the PR title, body, discussion, reviews and inline comments with URLs and collection time, locates security and threat-model documents, exports the sources with a snapshot script, and seals a preparation record.

What You Get

  • A sealed prepare.json with boundary explanations, frozen context and source artifacts
  • Frozen S0 and S1 source exports in a new directory
  • Collected PR or commit discussion with URLs, collection time, limits and lifecycle
  • Paths to security or threat-model documents in both states, or notes on search limits
  • A short report back with the selected S0 and S1 and collection limitations

Common Use Cases

  • Starting a regression investigation on a merged pull request
  • Freezing an unmerged PR at its current head before review
  • Preparing a commit range for a release audit

What Makes It Unique

A merged status does not define the change boundary here. The skill spells out how to pin S0 and S1 for each way a PR can land, blocks on ambiguous ranges, and records what could not be collected so that later stages know the limits.

Under The Hood

The agent reads SKILL.md plus a regression definition and a file contract. It calls PR and commit APIs, paginated, to collect discussion, runs a bundled Python snapshot script through uv with Git to export both states into a new directory, and writes and seals a JSON record there. It reads project guidance files once the target commit is known and treats them as data. It runs no target code and makes no changes to the repository.

Join founders building with open source

Opinionated takes, migration guides, cost-saving tips, and insights from the open source ecosystem.

Subscribe on SubstackJoin 750+ subscribers