The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes several shell commands to automate documentation cleanup, including python3 for script execution (tidy.py, wiki-health.py) and git for moving files (git mv) and checking repository status (git status). It also invokes a bash script harness/tests/tidy-test.sh for verification.
[INDIRECT_PROMPT_INJECTION]: This skill presents an attack surface for indirect prompt injection because it requires the agent to ingest and analyze potentially untrusted data from draft markdown files and HTML renders. Malicious instructions embedded in these files could attempt to manipulate the agent during the "promote" phase where it decides which content is valuable enough to move to the wiki.
Ingestion points: The agent reads data from wiki/sources/draft/*.md and llmwiki/html/*.html to identify decisions, lessons, or patterns.
Boundary markers: There are no instructions to use delimiters or defensive prompting when reading these external files.
Capability inventory: The agent can execute shell scripts, perform git operations, and write new markdown files to the wiki (llmwiki/wiki/sources/adr/ or llmwiki/wiki/concepts/).
Sanitization: The skill lacks explicit sanitization or validation steps for the content being read and promoted to ensure it does not contain malicious payloads or prompt injection attempts.

tidy