The Agent Skills Directory

[DATA_EXFILTRATION]: The skill explicitly instructs the agent to search for and read sensitive information, including environment variables and DSN (Data Source Name) configurations for error tracking platforms like Sentry across various project files.
[COMMAND_EXECUTION]: Utilizes Bash and Grep to perform automated reconnaissance on the local filesystem, identifying service definitions, build targets, and deployment configurations.
[DATA_EXFILTRATION]: The skill requests WebFetch and WebSearch permissions and specifies a delivery mechanism (/atlas-report) that involves sending collected audit data to an external reporting endpoint. This creates a data flow from sensitive local project configurations to external infrastructure.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it ingests large amounts of untrusted data from the project environment.
Ingestion points: Files including package.json, go.mod, requirements.txt, pyproject.toml, Cargo.toml, Dockerfile, docker-compose.yml, Kubernetes manifests, and log files (SKILL.md).
Boundary markers: None identified; the skill does not use delimiters or instructions to ignore embedded commands in the files it reads.
Capability inventory: The skill has access to Bash, Read, Glob, Grep, WebFetch, and WebSearch tools.
Sanitization: There are no instructions for validating or escaping content read from the filesystem before processing or reporting.

vigil-recon