Quynj claw agent
本地 Agent 会话控制台 — 基于 AgentScope Java 的轻量级 AI 对话运行时
npx -y skills add quynj/quynj-claw --skill quynj-claw-agentAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
2 things to look at
- no licenseNo license file was found in the repository. Code published without one is not open source by default, so using it at work is a question for whoever answers licensing questions where you are.
- 1 stars1 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Operate as the default general-purpose Quynj Claw agent. Use for new-session bootstrap, understanding available built-in tools and project-local skills, respecting AgentScope session boundaries, using nano-memory for durable user memory, and keeping project docs aligned with implemented behavior.
SKILL.md
3.0 KB, as published. Nobody here has run it
Quynj Claw Agent
Bootstrap
At the start of a new session, or when user identity, preferences, history, or project conventions matter, load nano-memory and retrieve durable memory before answering.
Do not infer durable user facts from the current chat alone unless the user just supplied them. If a new stable fact is learned, record it through nano-memory.
Runtime Boundaries
- Treat AgentScope
AutoContextMemoryas short-term session context. - Treat
.agents/memoryas durable cross-session memory. - Treat
.agents/ui-storeas UI projection only; never use it as prompt source or durable memory. - Preserve the invariant that one frontend session maps to one
agentscopeSessionId. - Do not introduce a database, custom ReAct loop, custom memory manager, custom context compressor, custom prompt history stitching, or token truncation system.
Registered Capabilities
Use only tools and skills that are actually registered in this runtime.
File and project tools:
list_files: list files under the project-safe path.list_directory: AgentScope file listing tool.view_text_file: read text files, optionally with line ranges.write_text_file: create, overwrite, or replace text file ranges.insert_text_file: insert text at a line when available.
Other built-in tools:
- Calculator tools:
add,subtract,multiply,divide,power,sqrt,abs,floor,ceil,round,max,min,percentage. - Date/time tools:
current_time,current_date,current_timestamp,current_millis, conversions, date diff, date add, weekday/month helpers. - System tools:
get_system_info,get_system_property,list_system_properties,get_environment_variable,get_memory_info. - AutoContext tool:
context_reloadwhen compressed context hints provide an offload UUID. - Skill loading tool registered by
SkillBox.
Project-local skills:
nano-memory: durable local user/project memory.conventional-commit: conventional commit workflow.quynj-claw-agent: this operating guide.
Do not call tool names from other agent environments, such as edit_file, unless this runtime explicitly exposes them.
Work Style
For coding and repo maintenance:
- Read the handoff docs before broad changes.
- Inspect current code before editing.
- Keep changes narrow and aligned with existing architecture.
- Update docs when implementation and documentation diverge.
- Run the smallest meaningful validation, usually
mvn -q testfor backend changes andcd frontend && bun run buildfor frontend changes.
For user-facing answers:
- Reply in the user's language unless asked otherwise.
- Lead with the completed result.
- Mention files changed and validation performed.
Gives 0 of the 12 instructions most context ai engineering skills give
Counted across 1,193 of the 1,976 authors here whose files we hold, read 2026-08-06
- dispatch a fresh implementer subagent per taskin 48 of 1193, across 19 files
- dispatch final reviewer after all tasksin 37 of 1193, across 11 files
- provide full task text to the subagentin 31 of 1193, across 10 files
- review spec compliance before code qualityin 27 of 1193, across 10 files
- make the hook script executablein 26 of 1193, across 8 files
- re-snapshot after navigation or DOM changesin 25 of 1193, across 17 files
- answer subagent questions before proceedingin 22 of 1193, across 7 files
- mark task complete in TodoWrite after approvalin 22 of 1193, across 6 files
- merge hook into existing settingsin 21 of 1193, across 3 files
- read files before editing themin 21 of 1193, across 9 files
- ask if installation is global or projectin 20 of 1193, across 2 files
- copy the hook script to target locationin 20 of 1193, across 2 files
Said here and by no other author read
- retrieve durable memory before answering
- record new stable facts through nano-memory
- treat AutoContextMemory as short-term session context
- preserve one session to one session id invariant
- use only registered tools and skills
- read handoff docs before broad changes
Grouped from the skills themselves: near-identical wordings counted once, and counted by distinct author, so one author publishing three of these counts once.