aws-infra
Verdict safe, but 2 findings (2 medium), mainly data exfiltration, unauthorized tool use. Most recent scan 2026-02-05.
The aws-infra skill is a moderately secure AWS management tool with good safety intentions but some implementation gaps. The skill demonstrates security awareness through explicit safety rules (read-only by default, confirmation for destructive actions, secret protection), but lacks technical enforcement mechanisms. Primary concerns include: (1) potential credential exposure through AWS CLI output without filtering, (2) missing allowed-tools declaration leaving tool usage unrestricted, and (3) access to AWS configuration files without validation guidance. The skill does not contain malicious code, data exfiltration attempts, or prompt injection. The safety rules are well-intentioned but rely on agent compliance rather than technical controls. Missing metadata fields reduce transparency. Overall, this skill is suitable for use with caution and would benefit from adding technical enforcement layers to match its stated safety policies.
Scanner's overall assessment (LLM analyzer)
Second scanner: NVIDIA SkillSpector
Independent static run, no language model, over the same files as the Cisco scan above. How the two scanners compare across the whole corpus: Two scanners, one corpus.
What this skill does
Chat-based AWS infrastructure assistance using AWS CLI and console context. Use for querying, auditing, and monitoring AWS resources (EC2, S3, IAM, Lambda, ECS/EKS, RDS, CloudWatch, billing, etc.), and for proposing safe changes with explicit confirmation before any write/destructive action.
- Package
- 3 files, 4.6 KB
- SKILL.md
- 233 words, 8 sections
- Code
- None. Instructions only.
SKILL.md covers: AWS Infra · Overview · Quick Start · Safety Rules (must follow) · Task Guide (common requests) · Region & Profile Handling · References · Assets
Findings
MEDIUM Potential Credential Exposure via AWS CLI Output Data Exfiltrationllm
The skill instructs the agent to run 'aws sts get-caller-identity' and other AWS CLI commands that may output sensitive information including account IDs, user ARNs, and access key IDs. While the safety rules state 'Never reveal or log secrets (access keys, session tokens)', the skill provides no technical enforcement mechanism to prevent the agent from displaying this information to the user or in logs. The agent will have access to AWS credentials from the environment (~/.aws/config, AWS_PROFILE, AWS_REGION) and could inadvertently expose credential metadata.
SKILL.md:12
Quick Start section: 'aws sts get-caller-identity' - This command outputs UserId, Account, and Arn which may contain sensitive identity information. Safety Rules section states 'Never reveal or log secrets' but provides no code-level enforcement.MEDIUM Unrestricted AWS CLI Tool Access Without allowed-tools Declaration Unauthorized Tool Usellm
The skill does not declare 'allowed-tools' in its YAML manifest, leaving tool usage unrestricted. The skill instructions direct the agent to execute arbitrary AWS CLI commands via Bash, including potentially destructive operations (delete, terminate, modify, scale). While the instructions include safety rules requiring confirmation, there is no technical enforcement at the manifest level to restrict which agent tools can be used. This creates a gap between stated safety policy and technical controls.
SKILL.md
YAML manifest missing 'allowed-tools' field. Instructions direct use of AWS CLI commands including destructive operations: 'aws s3 rm', 'aws ecs update-service', 'aws autoscaling update-auto-scaling-group'. No technical restriction prevents the agent from executing these commands via Bash tool.What is in the package
| File | Kind | Size |
|---|---|---|
SKILL.md | doc | 1.9 KB |
assets/icon.svg | data | 1020 B |
references/aws-cli-queries.md | doc | 1.7 KB |
Environment variables referenced
AWS_PROFILE AWS_REGION
Package contents recorded 2026-10-22. The scan above ran against the version current at scan time; the two can differ if the skill has since been updated.
What we checked
- Static analyzer. Pattern rules over SKILL.md and bundled files: known-bad commands, exfiltration URLs, secrets, obfuscation.
- Behavioral analyzer. Checks what the skill asks the agent to do against what its description claims it does.
- Llm analyzer. A language model reads the skill and reports intent-level issues the rules cannot express.
- Trigger analyzer. Looks for instructions that fire on conditions the user did not ask for.
- Meta analyzer. A second model pass that validates, deduplicates and re-scores the other analyzers' findings.
What we did not check
- Runtime behaviour. The skill was not executed; findings come from reading its files.
- Anything fetched at install or run time, including remote scripts, package dependencies, and URLs the skill tells the agent to open.
- Content the analyzers cannot read: images, binaries, and some non-English text.
- Known scanner bypass classes. See the methodology and limitations page.
Sources
- Repository: https://clawhub.ai/skills/aws-infra
- Raw scan result: aws-infra-scan.json
- Badge for your README:
- Request a re-scan