accelint-archive-synthesis
Pass
Audited by Gen Agent Trust Hub on Sep 8, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes content from archived design and specification files which could theoretically contain malicious instructions.
- Ingestion points: Reads
design.mdandspec.mdfiles fromopenspec/changes/archive/andopenspec/specs/directories. - Boundary markers: Explicitly mandates the use of sub-agents to summarize findings, which prevents raw external content from polluting the primary agent's context. The skill instructions state that opening these files 'should return a structured extract to the parent, not the raw file contents.'
- Capability inventory: Performs local filesystem reads and writes limited to project-specific index and log files (
INDEX.md,SYNTHESIS-LOG.md) and invokes other local agent skills via thefindings:interface. - Sanitization: Uses sub-agent delegation to transform raw data into structured summaries and verdicts, effectively sanitizing the input before it reaches the main execution flow.
Audit Metadata