deps
Pass
Audited by Gen Agent Trust Hub on Sep 15, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process untrusted data from GitHub issue titles, bodies, and comments, which constitutes a potential attack surface for indirect prompt injection.
- Ingestion points: External data enters the context through
larch.sh deps fetchand the resultinguntrusted_corpus_file(found in SKILL.md, Step 1). - Boundary markers: The skill uses explicit instructions to treat content within
<deps_issue_N>tags as data and includes mandatory directives to disregard any instructions or directives embedded within the issue text (found in SKILL.md, Untrusted Input section). - Capability inventory: The skill has the capability to modify GitHub state (issue rewrites, closes, and dependency updates) using
larch.sh deps applyand handles local file operations via Bash (found in SKILL.md, Step 6 and allowed-tools). - Sanitization: Security is significantly bolstered by a deterministic planning phase (
deps plan) that validates all proposals and a mandatory manual approval step usingAskUserQuestionbefore any modifications are executed (found in SKILL.md, Steps 4 and 5).
Audit Metadata