agentsclimarketplace
Sign in

Skill2 test

Skill blackblue-labs/skill2/skills/skill2-test

Use when checking whether a skill activates, avoids false activation, improves outcomes, or works in an isolated agent environment.From its SKILL.md

Install
npx -y skills add blackblue-labs/skill2 --skill skill2-test

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

SKILL.md

2.2 KB, 422 tokens by cl100k_base, as published. Nobody here has run it

Test Agent Skills

Measure behavior without inheriting current chat, global memory, or unrelated Skills.

Ownership

  • Test owns live activation, outcome, routing, and baseline evidence.
  • Audit owns static structure and trigger-overlap review.
  • Create owns Skill changes; Test never rewrites a failing Skill automatically.

Layers

LayerInstalled SkillsMeasures
Target-onlyTarget Skill onlyActivation, outcome, baseline uplift
PackCandidate sibling SkillsRouting and false activation

Use target-only for core outcomes. Use pack for adjacent and unrelated routing cases.

Cases

  • Core positive: target should activate and produce required artifact or answer.
  • Paraphrase: same intent, different wording.
  • Adjacent: sibling should own request.
  • Unrelated: no package Skill should activate.
  • Assertions: inspect files, output, commands, or explicit events.

Default one trial. Add repetitions only for known nondeterminism or regression confidence.

Isolation

  • Fresh session, temporary home, temporary workspace.
  • Install only layer-required Skills.
  • Minimal authentication/configuration; no user memory or chat history.
  • Guard host-home reads and writes.
  • Adapter owns harness-specific install and event parsing.
  • Save raw events, final output, workspace, version, and Skill hash locally.

Verdicts

  • pass: activation and outcome assertions pass.
  • fail: deterministic assertion fails.
  • inconclusive: runner or evidence cannot support claim.
  • Baseline also passes: no demonstrated deterministic uplift.
  • Exact isolated SKILL.md read is medium-confidence activation; explicit activation event is high confidence.

Fake runner tests validate adapter plumbing, not provider behavior. Never claim unrun results.

uv run --script <skill-dir>/scripts/run -- test skills/<name> --agent <agent> --cases cases/<name>.yaml --baseline
uv run --script <skill-dir>/scripts/run -- test skills/<name> --agent <agent> --cases cases/<name>.yaml --pack

What ships with it: 10 files

72.4 KB alongside SKILL.md, 9 of them executable

scripts/

Gives 0 of the 12 instructions most test skills give in 422 tokens

Counted across 1,201 of the 2,096 authors here whose files we hold, read 2026-09-06

  • Write a failing test before writing codein 43 of 1201, across 36 files
  • Run the full test suitein 36 of 1201, across 35 files
  • Test only one variable per experimentin 34 of 1201, across 17 files
  • Read product marketing context before asking questionsin 34 of 1201, across 14 files
  • Mock external dependenciesin 34 of 1201, across 30 files
  • Define primary, secondary, and guardrail metricsin 33 of 1201, across 16 files
  • Pre-determine sample size before startingin 31 of 1201, across 14 files
  • Test behavior rather than implementationin 31 of 1201, across 29 files
  • Formulate a hypothesis before designing a testin 30 of 1201, across 13 files
  • Document every test hypothesis, variant, and resultin 29 of 1201, across 11 files
  • Use descriptive test function namesin 25 of 1201, across 21 files
  • Commit to the methodology without stopping earlyin 24 of 1201, across 8 files

Said here and by no other author read

  • Measure behavior in isolated agent environments
  • Use target-only layer for core outcomes
  • Use pack layer for routing cases
  • Perform one trial by default
  • Save raw events and final output locally
  • Guard host home reads and writes

Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once. Length counted with cl100k_base; the agent that loads this file may tokenize it differently.

Keep looking

Skills are one crate of 325,949. Ordering is by how many stacks a row turns up in, so the top of any crate is what has actually been picked rather than what has the most stars.