quality-tool-picker
Skill1 starsChanged 21 days agoMITView on GitHub
Research and compare the current options for one tooling slot in a project (formatter or style guide, linter or lint preset, type checker, test runner, coverage tool, mutation tester, HTTP client, retry library, logger, and similar), using live web research with cited sources, then help the user choose and record the decision. Checks the org's toolkit first. Use when the user or the code-quality skill needs to choose between tools, e.g. "which Java formatter should we use", "Spotless with google-java-format or palantir?", "compare ruff and black", "pick a retry library for Go". Not for configuring a tool that has already been chosen.
In this repositoryOne other file it gives its agents.
- code-qualitySkill
The file
We split this skill into the rules it gives an agent
8 rules. About 0 are common, 5 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈5
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 42
Start from references/catalog.md (a seed list, which may be out of date), then search the web for current alternatives and recent …
- Distinctive: line 45
Keep 2 to 4 serious candidates.
- Distinctive: line 49
Prefer primary sources: the official repo, release notes, documentation and the package registry page.
- Distinctive: line 63
Write down where each fact came from and the date you checked.
- Distinctive: line 88
Keep it to what matters for this decision.
- Distinctive: line 94
Ask the user to choose, listing the recommended option first.
- Very specific: line 99
Write a short decision record so the choice isn't reopened every few months, and so future agents follow it.
- Very specific: line 123
If the code-quality skill sent you here, return the choice so it can configure the tool using its language reference.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Quality tool picker
Picks between tools that fill the same slot, such as google-java-format vs palantir-java-format vs the Eclipse formatter for Java formatting. The decision belongs to the user or the org. This skill does the research, lays out the trade-offs with sources, recommends one option, and records whatever the user decides.
Tool landscapes change quickly: projects get abandoned, forked or replaced, and a new tool can overtake the old default within a year. Never answer from memory alone. Check the current state on the web and cite what you found.
1. Pin down the slot and the constraints
Collect these from the project and the user before researching:
- Slot: exactly what the tool must do (for example "Java formatting, enforced in CI", not "Java quality").
- Stack: language and version, build tool, framework, CI provider, editors in use.
- Existing setup: what the project already uses for this slot, and what migrating away would cost.
- Org toolkit: shared configs, parent POMs, convention plugins, CI templates, an internal library. If the org mandates a tool for this slot, stop and report that choice. There is nothing to pick.
- Must-haves and deal-breakers: licence, offline use, auto-fix, speed, IDE support, compatibility with other tools (for example a formatter that doesn't fight the linter).
2. Build the candidate list
Start from references/catalog.md (a seed list, which may be out of date), then search
the web for current alternatives and recent comparisons, for example "<slot> <language>
<current year>", "<tool A> vs <tool B>", and "<tool> deprecated" or "<tool> alternative".
Keep 2 to 4 serious candidates. Include "keep what we have" when something already exists.
3. Research each candidate
Prefer primary sources: the official repo, release notes, documentation and the package registry page. Use blog posts and comparisons for context only, never as the only source for a fact. For each candidate, find:
| Aspect | What to check |
| Maintenance | latest release and its date, release cadence, recent commits, how issues get answered, whether it's deprecated or has a successor |
| Adoption | registry downloads or well-known users, where the ecosystem publishes them |
| Compatibility | supports the project's language and runtime versions, build tool and framework |
| Fit | covers the slot completely; overlaps or conflicts with tools already chosen |
| Cost | setup effort, migration effort (for a formatter: one large reformat commit), CI time |
| Developer experience | IDE and editor integration, auto-fix, quality of error messages |
| Licence | compatible with the project |
Write down where each fact came from and the date you checked. If you can't verify something, say "unknown". Never estimate numbers such as download counts or release dates.
4. Compare and recommend
Present a short comparison and one recommendation:
## <Slot>: options for <project>
| | <Option A> | <Option B> | <Option C> |
|---|---|---|---|
| Latest release | 1.2.3 (2026-09-14) | ... | ... |
| Maintenance | active, monthly releases | ... | ... |
| Fits our stack | yes (JDK 21, Gradle 9) | ... | ... |
| Main trade-off | ... | ... | ... |
| Migration cost | ... | ... | ... |
**Recommendation: <Option>**, because <reasons tied to the constraints in step 1>.
Runner-up: <Option>, if <condition>.
Sources (checked <date>): <links>Keep it to what matters for this decision. If the options are effectively equivalent, say so and recommend the one that's cheapest to adopt, which is often the one already in use or the one the wider ecosystem defaults to.
5. Let the user decide
Ask the user to choose, listing the recommended option first. If they choose differently, that's their call: record it without arguing further.
6. Record the decision
Write a short decision record so the choice isn't reopened every few months, and so
future agents follow it. Use the project's existing convention if it has one, otherwise
docs/adr/NNNN-<slot>.md:
# NNNN. <Slot>: <chosen tool>
Date: <date>. Status: accepted. Review by: <date, about a year out>.
## Context
<slot, constraints, org toolkit status>
## Options considered
<one line each, with the main trade-off>
## Decision
<tool and version, plus the key configuration choices, such as a style preset>
## Consequences
<migration steps, CI changes, what becomes harder>
Sources (checked <date>): <links>If the code-quality skill sent you here, return the choice so it can configure the tool
using its language reference.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

