telegram-channel-parser
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill's primary function is to ingest and display untrusted content from public Telegram channels. This content is then analyzed by the AI agent to produce summaries or digests, creating a surface for indirect prompt injection.
- Ingestion points: Untrusted data enters the system through
scripts/fetch_posts.sh, which scrapes HTML content from thet.me/s/web preview service. - Boundary markers: While the skill organizes data into structured formats like TSV and JSON, it does not wrap the extracted message text in explicit delimiters or system-level instructions to ignore embedded commands before the agent reads them.
- Capability inventory: The skill possesses the ability to execute local shell scripts, perform network requests to Telegram's domain, and write to the local filesystem (e.g., in
/home/claude/telegram-channel-parser). - Sanitization: The
scripts/parse_tg_posts.awkutility includes aclean_textfunction that stripsspantags and potentially dangerous HTML attributes (likestyleanddir), but it retains the core text content of the posts which is where injection instructions would be located.
Audit Metadata