The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes local bash commands to facilitate research tasks, such as running a training module located in mental-gym/.venv/bin/mental-gym. It also utilizes Git for version control, executing git status, git add, git commit, and git push to manage the lifecycle of research data.
[DATA_EXFILTRATION]: At the end of a session, the skill performs a git push origin main. This operation transmits the research repository's contents, including logs, state files, and wiki updates, to a remote server. This is a standard synchronization step for the intended workflow but constitutes an external network transfer of project data.
[PROMPT_INJECTION]: The skill exhibits a surface for indirect prompt injection as it ingests and processes content from files like IDEAS.md and wiki pages, which may contain data derived from external research sources.
Ingestion points: Files including .claude/research-state.yaml, events.jsonl, IDEAS.md, and markdown files in the wiki/ directory.
Boundary markers: None explicitly defined to separate the instructions from the data read from these files.
Capability inventory: The agent has access to powerful tools including Bash, Edit, Write, and network-enabled tools like WebSearch and WebFetch.
Sanitization: There is no evidence of content validation or escaping before the data is processed into the agent's context.

research-session