The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses various shell commands to establish context and inventory the scope of a review, including git status, git rev-parse, rg, ls, cargo metadata, git diff, and make.
[DYNAMIC_EXECUTION]: In Step 5, the skill executes a local Python script at plugins/imbue/scripts/citation_verifier.py. This script processes a generated JSON file (.review/findings.json) to verify that the reported findings are accurately grounded in the source code.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process untrusted external data from the repository, creating a potential injection surface.
Ingestion points: The skill reads source files, configurations, documentation, and specifications in Step 2 and Step 3.
Boundary markers: There are no explicit instructions or delimiters defined to prevent the agent from obeying instructions embedded within the files being reviewed.
Capability inventory: The skill can execute shell commands, write to local files (.review/findings.json, docs/lessons-learned.md), and run a Python verification script.
Sanitization: While the citation_verifier.py script checks for the existence of code anchors, there is no mention of sanitizing or escaping the content of these anchors or findings before they are written to disk or processed by the agent.

review-core