v1.0.7

Anthropic Usage

BenignClawScan verdict for this skill. Analyzed May 1, 2026, 7:38 AM.

Analysis

This skill is coherent for checking Anthropic usage, but users should notice it requires an Anthropic Admin API key and runs a local shell script that contacts Anthropic.

GuidanceThis appears safe for its stated purpose if you intend to check Anthropic organization usage. Before installing, be comfortable providing an Anthropic Admin API key, expect calls to api.anthropic.com and a possible fetch of anthropic.com/pricing, and protect or rotate the key if it is ever exposed.

Findings (4)

Artifact-based informational review of SKILL.md, metadata, install specs, static scan signals, and capability signals. ClawScan does not execute the skill or run runtime probes.

Abnormal behavior control

Checks for instructions or behavior that redirect the agent, misuse tools, execute unexpected code, cascade across systems, exploit user trust, or continue outside the intended task.

Unexpected Code Execution

SeverityLowConfidenceHighStatusNote

SKILL.md

Once the key exists, run `scripts/usage.sh` with the appropriate flags

The skill operates by running an included shell script. This is disclosed and central to the purpose, but it is still local command execution.

User impactWhen invoked, the agent may run the included script locally to make Anthropic API requests and format the returned usage data.

RecommendationReview the script if desired and ensure you are comfortable with local shell execution using curl and jq before enabling the skill.

Tool Misuse and Exploitation

SeverityInfoConfidenceHighStatusNote

SKILL.md

use your `web_fetch` tool to fetch `https://www.anthropic.com/pricing` ... This fetch is intentional and scoped to `anthropic.com`.

The skill asks the agent to perform an external web fetch for pricing, but the target is disclosed and limited to Anthropic's pricing page.

User impactCost estimates may trigger an additional request to Anthropic's public pricing page.

RecommendationExpect the pricing-page fetch when asking for costs; verify estimates if billing accuracy is important.

Human-Agent Trust Exploitation

SeverityLowConfidenceHighStatusNote

README.md

Your key is stored in `~/.openclaw/openclaw.json` and never leaves your machine.

This privacy wording is over-absolute: the script necessarily sends the key to Anthropic's API as an authentication header, although no evidence shows it being sent to unrelated endpoints.

User impactA user could misunderstand the credential flow and think the key is never transmitted at all.

RecommendationInterpret this as no unrelated sharing being shown; the key is still sent to Anthropic to authenticate API requests.

Permission boundary

Checks whether tool use, credentials, dependencies, identity, account access, or inter-agent boundaries are broader than the stated purpose.

Identity and Privilege Abuse

SeverityMediumConfidenceHighStatusNote

SKILL.md

primaryEnv: ANTHROPIC_ADMIN_API_KEY

The skill requires an Anthropic Admin API key, which is expected for organization usage reporting but is a high-value credential.

User impactAnyone using this skill must trust it with an Anthropic organization Admin API key and the ability to read usage-report data.

RecommendationUse the OpenClaw key field, avoid sharing the key in chat, rotate the key if exposed, and install only if you are comfortable granting usage-report access.