curated-skillsCommunity·10 Agent Skills Included

LinguaClaw

Natural-language harnesses for autonomous software issue resolution

Runs autonomous software-engineering workflows that reproduce bugs, patch source code, and verify fixes through reusable markdown harnesses. Coordinates subagents, multi-candidate patch search, verification stages, and self-improving retry loops without hard-coded controller logic. Eliminates brittle framework glue code so teams can compose, audit, and benchmark repair workflows faster.

43|2

↗

npx skills add curated-skills/LinguaClaw --all -g -y

All Skills in This Repository (10)

Pure Emerald Level Indicators

📦 In Repo

curated-skills

agent-harness-runtime-charter

Define runtime policy for delegating child agent work and reconstructing executable call graphs.

swe-agent

Reproduce repository failures and generate minimal verified git patches.

live-swe-agent

Iteratively debug, patch, and verify software issues within a repository.

trae-agent

Select the best repository patch for a GitHub issue via candidate search and majority voting.

file-backed-state-module

Externalize agent state into durable files with path-addressable handoffs.

dynamic-subagent-orchestration-module

Select orchestration topologies for supervised parallel or staged subagent execution.

verifier-module

Verify candidate answers against original problems with verdict labels and reports.

self-evolution-module

Orchestrate reflection-driven retry loops that evolve prompts, tools, and workflows.

evidence-protocol-module

Enforce citation-backed evidence documentation before releasing final answers.

multi-candidate-search-module

Execute multi-candidate search with budgeted diversification, pruning, and evidence-based selection.

Community

Advanced

Frequently Asked Questions

FAQPage Schema

How to install LinguaClaw?▼

Run `npx skills add curated-skills/LinguaClaw --all -g -y` in your terminal to install all skills in this suite globally.

What does LinguaClaw do?▼

It provides a thin runtime plus reusable natural-language harnesses that let an agent reproduce repository bugs, write minimal patches, and verify fixes with tests.

How does multi-candidate patch search work?▼

The TRAE-style harness generates several independent candidate fixes, prunes weak ones, and selects the strongest patch through majority voting and final revalidation.

Can LinguaClaw verify its own fixes?▼

Yes. The verifier module independently audits each candidate against the original problem and only approves fixes that pass concrete checks.

Do I need to write code to use these harnesses?▼

No. Harness behavior is defined in plain markdown files, so you can compose and customize workflows by editing natural-language instructions.