The Agent Skills Directory

[SAFE]: The skill performs authorized diagnostic tasks by reading local configuration and project files to provide context usage metrics. These actions are consistent with its stated purpose and do not involve unauthorized data access or external transmission.
[PROMPT_INJECTION]: The skill ingests untrusted data from project files, creating a potential surface for indirect prompt injection. Ingestion points: Reads content from agents/*.md, skills/*/SKILL.md, rules/**/*.md, .mcp.json, and CLAUDE.md. Boundary markers: None identified; the skill directly processes content for token estimation. Capability inventory: The skill provides reporting and analysis but lacks high-risk capabilities like network access, file writing, or arbitrary command execution. Sanitization: No content sanitization or filtering is implemented, though the lack of exploitable capabilities mitigates the associated risk.

context-budget