tldr-expert
Pass
Audited by Gen Agent Trust Hub on Oct 17, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process entire repositories to create context bundles for LLMs. This functionality creates an attack surface where an agent might process and follow malicious instructions hidden within the source code files or documentation of a project it is analyzing.
- Ingestion points: Processes local repository files using
repomixandgitingestto generate digests and bundles as described in SKILL.md and references/2-repomix-gitingest-mastery.md. - Boundary markers: Recommends the use of XML tagging to clearly distinguish between file content and instructions, providing structural delimiters for the agent (noted in references/1-context-engineering-patterns.md).
- Capability inventory: The skill utilizes tools for reading, indexing, and structural analysis of code (llm-tldr, Repomix, Gitingest); it does not include instructions for executing the code or making unauthorized network requests.
- Sanitization: Explicitly includes a security section in SKILL.md advising the use of
secretlintwithin the Repomix workflow to ensure that sensitive data like API keys, PII, and internal metadata are filtered out before context is provided to the agent.
Audit Metadata