The Agent Skills Directory

[PROMPT_INJECTION]: The skill explicitly instructs the agent to "Run fully automatically — no user prompts or confirmations at any point" within the prompt builder in scripts/generate-docs.sh and in the invocation section of SKILL.md. This bypasses standard user interaction loops and interaction constraints, significantly reducing safety oversight.
[PROMPT_INJECTION]: The skill exhibits a vulnerability surface for indirect prompt injection by ingesting untrusted data from the codebase being documented. 1. Ingestion points: The research agents, specifically Agent 5 in SKILL.md and RESEARCH-PROMPTS.md, are tasked with mining git logs, merge commits, plan files, and code comments. 2. Boundary markers: The instructions fail to provide boundary markers or directives to ignore instructions embedded within the ingested codebase data. 3. Capability inventory: The skill utilizes Write, Edit, and Bash tools, which could be abused if the agent executes instructions found in the scanned code. 4. Sanitization: There is no evidence of sanitization, validation, or escaping of the content read from the codebase before it is processed by the AI.
[COMMAND_EXECUTION]: The skill uses the Bash tool to run commands including git, tree, ls, and find for repository discovery and analysis. These commands are executed without user confirmation due to the autonomy instructions in the skill definition.

superdocs