freema/CodeForge

AI-powered code review & sessions for GitHub and GitLab — multi-CLI, auto-knowledge, review posting

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
session_typeSession type: pr_review, code_review, knowledge_update, customnopr_review
promptCustom prompt (required for session_type=custom, optional for reviews)no
cliAI CLI tool: claude-code or codexnoclaude-code
modelAI model override (e.g., claude-sonnet-4-20250514)no
api_keyAI provider API key (overrides ANTHROPIC_API_KEY / OPENAI_API_KEY)no
provider_tokenGitHub/GitLab token for PR operations (defaults to GITHUB_TOKEN)no
mcp_configMCP configuration JSON string or path to .mcp.jsonno
post_commentsPost review results as PR/MR commentsnotrue
output_formatOutput format: json, markdown, textnojson
max_turnsMaximum AI conversation turnsno
allowed_toolsComma-separated tool allowlist for Claude Codeno
pr_numberPR number to review (auto-detected from event, use for issue_comment trigger)no
fail_on_request_changesExit with code 1 when verdict is request_changesnofalse
namedescription
verdictReview verdict: approve, request_changes, or comment
scoreReview score (1-10)
issues_countNumber of issues found
reviewFull review result as JSON
input_tokensInput tokens consumed
output_tokensOutput tokens consumed
outputRaw CLI output