← Back to Skills Marketplace
107
Downloads
0
Stars
0
Active Installs
1
Versions
Install in OpenClaw
/install multi-capability-bug-closure
Description
统一调用源码、数据库、服务器日志、软件平台查询等多能力体系进行 BUG 定位与闭环论证。用于用户要求“必须基于真实数据给出结论与证据链”，而非仅代码静态分析的场景。
Usage Guidance
This Skill is coherent in purpose (it truly needs code, DB, log, and platform access to do what it claims) but it leaves important access details unspecified. Before installing or running it, confirm: (1) which connectors/credentials the agent will use (explicitly list databases, environments, and log paths); (2) provide read-only, least-privilege credentials or temporary tokens scoped to the minimum data needed (avoid using production admin keys); (3) restrict log and source access to the relevant service/component directories only; (4) require the agent to show a planned list of exact queries/paths it will read and ask for explicit user approval before executing them; (5) ensure any referenced third-party Skills (e.g., server-log-analysis) are trusted; (6) enable audit logging of the agent's actions and review outputs for sensitive data before allowing broader access. If you cannot safely provide narrowly scoped, read-only access, do not run the Skill with elevated or broad credentials.
Capability Analysis
Type: OpenClaw Skill
Name: multi-capability-bug-closure
Version: 1.0.0
The skill bundle defines a structured workflow for an AI agent to perform bug localization using source code, databases, and server logs. The instructions in SKILL.md are designed to ensure the agent provides evidence-based conclusions rather than just static analysis, and it includes explicit constraints to protect sensitive credentials and avoid faking data. No malicious code, data exfiltration, or harmful prompt injection was found.
Capability Assessment
Purpose & Capability
The name and description match the runtime instructions: the skill legitimately needs code, DB, server logs, and platform query capabilities to perform evidence-backed bug closure. However, the SKILL.md expects the agent to access those sensitive resources but the skill metadata declares no required environment variables, credentials, or config paths. That mismatch (implicit high-privilege access without declared requirements) is notable.
Instruction Scope
The SKILL.md explicitly directs the agent to read workspace source files, execute at least one read-only SQL on target databases, read server log configurations and target log directories/samples, and call platform Skills/documentation. These actions reach into sensitive areas (databases, logs, source code). The instructions do require not fabricating evidence and to avoid leaking creds, but they give broad discretion to access unspecified files/paths and other Skills — increasing the risk of unintended data exposure if connectors/permissions are not tightly scoped.
Install Mechanism
No install specification and no code files are present. As an instruction-only Skill, it does not download or install artifacts on disk, which reduces supply-chain/install risk.
Credentials
The Skill requires (in practice) database access, server log access, and the ability to read project source — all of which normally require credentials or config paths. Yet requires.env and required config paths are empty. Asking the agent to perform SQL queries and access logs without declaring or describing which credentials will be used is disproportionate and ambiguous. The SKILL.md does recommend using env/secret management, but it doesn't specify which secrets the platform must provide nor request read-only/least-privilege access.
Persistence & Privilege
always:false and no install steps are present; the Skill does not request persistent/system-wide presence or modification of other Skills. It can invoke other Skills (platform default), which is expected for this use case and is not by itself flagged here.
How to Use
- Make sure OpenClaw is installed (local or Docker)
- Run the install command in chat:
/install multi-capability-bug-closure - After installation, invoke the skill by name or use
/multi-capability-bug-closure - Provide required inputs per the skill's parameter spec and get structured output
Version History
v1.0.0
Initial release of multi-capability-bug-closure skill:
- Provides a standardized, multi-source approach for evidence-based bug localization and closure.
- Enforces mandatory use of real data from code, database, server logs, and software platform queries before drawing conclusions.
- Includes strict capability checks before execution; alerts users if any data source is unavailable and provides actionable recommendations.
- Follows a required workflow: issue structuring, server log and database forensics, code mechanism analysis, software platform validation, evidence chain merging, and actionable closure/reporting.
- Supplies a clear and replicable output template covering evidence, root cause, confidence rating, remediation steps, and verification criteria.
- Prohibits definitive conclusions without full data; lists missing evidence if capabilities are incomplete.
Metadata
Frequently Asked Questions
What is multi-capability-bug-closure?
统一调用源码、数据库、服务器日志、软件平台查询等多能力体系进行 BUG 定位与闭环论证。用于用户要求“必须基于真实数据给出结论与证据链”，而非仅代码静态分析的场景。 It is an AI Agent Skill for Claude Code / OpenClaw, with 107 downloads so far.
How do I install multi-capability-bug-closure?
Run "/install multi-capability-bug-closure" in the OpenClaw or Claude Code chat to install it in one step — no extra setup required.
Is multi-capability-bug-closure free?
Yes, multi-capability-bug-closure is completely free, licensed under MIT-0. You can download, install and use it at no cost.
Which platforms does multi-capability-bug-closure support?
multi-capability-bug-closure is cross-platform and runs anywhere OpenClaw / Claude Code is available (cross-platform).
Who created multi-capability-bug-closure?
It is built and maintained by hgvgfgvh (@hgvgfgvh); the current version is v1.0.0.
More Skills