show-and-tell
Pass
Audited by Gen Agent Trust Hub on Sep 29, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data from multiple external and internal sources to summarize daily work.
- Ingestion points: Slack activity (via Atlas MCP), Notion page content (via Notion MCP), and local agent session transcripts (
~/.claude/projects/*.jsonl). - Boundary markers: The skill does not define specific delimiters to isolate ingested data from instructions within the generated Markdown artifact.
- Capability inventory: The skill has the ability to execute shell commands (
bash) for running tests and builds, perform network requests (curl) for API verification, and take screenshots (chrome-devtools). - Sanitization: The skill contains explicit rules to redact PII and credentials before embedding them in evidence and requires the user to confirm the curated highlights before the agent proceeds with evidence collection.
- [COMMAND_EXECUTION]: The skill generates and executes shell commands such as
pnpm test,pnpm build, andcurlto provide evidence of work progress. These commands are dynamically determined based on the context found in logs, Slack messages, and Notion docs, which presents a risk of command injection if those sources contain malicious payloads.
Audit Metadata