jasonrve/TerraSec

Call an OpenAI-compatible LLM with a security-focused prompt, render a well-formatted Markdown report, and optionally upsert it as a PR comment.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
promptPrimary instruction for the LLM.yes
contextOptional additional context to include with the prompt.no""
llm_base_urlOpenAI-compatible LLM base URL.no""
llm_api_keyOpenAI-compatible LLM API key.no""
github_tokenGitHub token used to upsert PR comments.no${{ github.token }}
modelModel name to use.noopenai/gpt-4o-mini
pr_numberPull request number to comment on. Defaults to the current PR context when available.no""
post_commentWhether to upsert a PR comment.notrue
fail_on_findingsFail the step when the generated report contains findings or a non-pass verdict.nofalse
comment_markerStable HTML marker used to update one comment instead of spamming duplicates.no<!-- terrasec -->
dry_runSkip posting a PR comment and only emit outputs.nofalse
mock_response_fileOptional path to a file containing a raw OpenAI-style response payload for validation.no""
context_modeContext strategy to use. diff, full, hybrid, or agentic.nodiff
extra_context_pathsComma or newline separated list of file paths to always include.no""
max_file_charsMaximum characters to load per file.no12000
max_follow_up_roundsMaximum number of additional file-request rounds in agentic mode.no1
namedescription
verdictThe parsed verdict from the model.
finding_countNumber of findings in the report.
comment_bodyRendered Markdown comment body.
comment_urlURL of the upserted PR comment, if any.