Skip to content
SkillModel-led968 tokens in use

regression-hunt

The coordinator for regression campaigns: dispatch fresh agents through prepare, hunt, reproduce, replay, adjudicate and report, with sealed records.

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

Install command as written in the repository's README

Skill label

measured from its files

What installing regression-hunt costs

27tokens idle
968tokens 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

Nothing detected: no shell, network, file writes, git or credentials in its files.

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
  • Examples0/15
  • Output15/15
  • Size10/15
  • Resources0/15

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

Regression-hunt runs a campaign that investigates regressions in pull requests of any status or in commits, including local ones. The coordinator saves the change list, sets up a campaign directory outside the plugin, saves root guidance files from each target commit, and reserves each stage’s assignment through a registry script.

It then dispatches a fresh agent for each stage, checks and accepts each sealed record before updating the index and ledger, retries blocked work, and requests reports on confirmed findings, unresolved work and coverage gaps. Saved progress can be resumed.

What You Get

  • A campaign directory with scope.json, campaign.json, progress.md, a registry and a ledger
  • Revision-specific copies of root guidance files saved by change and target SHA
  • Sealed records from each stage, accepted by the registry before the index is updated
  • Reports covering confirmed findings, unresolved work and coverage gaps
  • A resume path that re-checks accepted records and reuses unchanged evidence

Common Use Cases

  • Running a regression review over a set of merged pull requests
  • Investigating commits in a local repository for introduced defects
  • Resuming a saved campaign after an interruption

What Makes It Unique

Confirmation needs separate reproduction, replay and adjudication agents, and dispatch, process exit or a completion message never counts as proof that work finished. The coordinator rejects the view that nothing reproduced means a clean review, and installation grants no permission for cloud spending, source export, publication or patching.

Under The Hood

The agent reads SKILL.md, a regression definition and an evidence file contract, and runs a bundled Python registry script to initialize the campaign, assign work, accept records, check them and retry stages. It fetches remote guidance files by SHA, reads local ones from Git objects, spawns fresh agents with full file contents in their prompts, and writes records, a ledger and progress notes into the campaign directory. It checks that stopped agents are terminated before a retry.

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