The Agent Skills Directory

[COMMAND_EXECUTION]: The skill directs the agent and user to run several setup and diagnostic scripts stored within the skill directory.
Evidence: References to init-environment.py, detect-backend.py, test-connection.py, and preflight-checks.sh in SKILL.md and references/troubleshooting.md.
[EXTERNAL_DOWNLOADS]: The skill requires the installation of external software packages and dependencies from the npm registry.
Evidence: Instructions for npm install -g @logseq/cli and npm install within the servers/logseq-mcp/ directory.
[REMOTE_CODE_EXECUTION]: The skill involves building a custom MCP server, which requires downloading unknown dependencies at runtime and executing build scripts.
Evidence: npm run build command for the Logseq MCP server in SKILL.md.
[PROMPT_INJECTION]: The skill facilitates reading data from external Logseq graphs, creating a surface for indirect prompt injection if the graph content contains malicious instructions.
Ingestion points: Data retrieved via logseq.Editor.getPage, logseq.DB.datascriptQuery (HTTP API), and logseq query (CLI).
Boundary markers: None detected; the skill does not specify delimiters to isolate graph data from the agent's control logic.
Capability inventory: The skill is granted Bash, Write, Edit, and Glob tools as per the YAML frontmatter in SKILL.md.
Sanitization: There is no evidence of content sanitization or instruction filtering for data ingested from Logseq.

connecting-to-logseq