The Agent Skills Directory

[COMMAND_EXECUTION]: The skill invokes the local utility engine runtime summary via the platform invocation mechanism to retrieve session statistics in JSON format.
[INDIRECT_PROMPT_INJECTION]: The skill synthesizes report prose from external audit shards and artefacts, which serves as a potential surface for indirect prompt injection. 1. Ingestion points: Markdown shards under the audit directory and recursive phase artefacts. 2. Boundary markers: None explicitly defined to isolate external prose during report generation. 3. Capability inventory: Terminal rendering only; no file writing or network exfiltration capabilities are present. 4. Sanitization: No explicit filtering or validation of the ingested narrative sources is performed.

aidlc-replay