The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes evidence.tsv (untrusted data generated from test runs) to generate reports and metrics. If the test output contains adversarial instructions, it could influence the agent's subsequent analysis or reporting behavior.
Ingestion points: The skill reads from qp/run/<new-ts>/evidence.tsv as input for analysis.
Boundary markers: No specific delimiters or "ignore instructions" markers are defined for the interpolation of test data into the final report.
Capability inventory: The skill uses Bash to execute processing scripts (qp_tools.sh) and Write to generate delta.tsv and report.md.
Sanitization: There is no evidence of sanitization or filtering of the content read from the evidence files before it is processed or reported.
[COMMAND_EXECUTION]: The skill executes a local shell script plugins/fh-qp/scripts/qp_tools.sh using the Bash tool. This script is used to compute surface-reach metrics from test data. This is a primary function of the skill and utilizes local project resources.

qp-regress