gadfly/.gitea/workflows/adversarial-review.yml

# Gadfly — agentic adversarial PR reviewer (https://gitea.stevedudenhoeffer.com/steve/gadfly).
#
# Runs the published Gadfly image (pinned to an immutable :sha- tag — act_runner
# caches :latest, and this build is what carries foreman provider-type support)
# as a specialist swarm and posts
# ONE consolidated review comment as gitea-actions. Advisory only — never blocks a
# merge. Gadfly reviewing its OWN PRs — dogfooding, full cloud fleet (9 cloud +
# the M5 Mac; M1 dropped as too slow), copied from mort's setup.

name: Adversarial Review (Gadfly)

on:
  pull_request:
    types: [opened, reopened, ready_for_review]
  issue_comment:
    types: [created]
  workflow_dispatch:
    inputs:
      pr_number:
        description: "PR number to review"
        required: true

permissions:
  contents: read
  issues: write
  pull-requests: write

concurrency:
  group: gadfly-${{ github.event.issue.number || github.event.pull_request.number || github.event.inputs.pr_number }}
  cancel-in-progress: true

jobs:
  review:
    # Security: only trusted users may trigger a secret-bearing run via a PR
    # comment (pull_request + workflow_dispatch are already trusted). Mirrors
    # GADFLY_ALLOWED_USERS, the in-container belt-and-suspenders check.
    if: >-
      github.event_name != 'issue_comment'
      || (github.event.issue.pull_request
          && (github.actor == 'steve'
              || github.actor == 'fizi'
              || github.actor == 'dazed'))
    runs-on: ubuntu-latest
    # Fleet (9 cloud + 1 local Mac/M5, all running concurrently) reviewing
    # every PR with the 3-lens suite — the slow local lane dominates wall time.
    timeout-minutes: 90
    steps:
      - uses: docker://gitea.stevedudenhoeffer.com/steve/gadfly:sha-d7f364d
        env:
          GITEA_API: ${{ github.server_url }}/api/v1/repos/${{ github.repository }}
          GITEA_TOKEN: ${{ secrets.GITEA_TOKEN }}
          OLLAMA_CLOUD_API_KEY: ${{ secrets.OLLAMA_CLOUD_API_KEY }}
          # Local Mac (M5), reached through its foreman queue (native Ollama on the
          # wire). Gadfly's GADFLY_ENDPOINT_* form with the "foreman" provider
          # type: GADFLY_ENDPOINT_M5 registers provider "m5", building a
          # foreman-preset Ollama client at the given URL. (M1 is dropped from
          # gadfly's swarm — too slow/low-signal — so its endpoint isn't mapped.)
          # Values
          # (host + token) live in gitea secrets, each of the form:
          #   foreman|https://<foreman-host>|<token>
          # (converted from the komodo LLM_* DSNs foreman://<token>@<host>).
          # REQUIRES a Gadfly image built with foreman provider-type support
          # (the GADFLY_ENDPOINT "foreman|..." type); on an older image the m1/m5
          # lanes error with "unknown provider foreman". The HTTPS-only LLM_*
          # foreman:// DSN is the alternative that needs no image rebuild.
          # NOTE: the Mac behind each foreman must still be awake/reachable; if a
          # box is offline, that model's comment shows an error and the others
          # still post. (Gitea secrets aren't auto-exposed — map each explicitly.)
          GADFLY_ENDPOINT_M5: ${{ secrets.GADFLY_ENDPOINT_M5 }}
          # Fleet: 9 cloud (3 at a time) + M5 Max — one consolidated comment each.
          # Matches mort's cloud set so the model-quality scoreboard is comparable
          # across both repos. NOTE: M1 Pro is intentionally dropped here (too slow
          # / low-signal for gadfly's own PRs); mort still runs it.
          GADFLY_MODELS: "minimax-m3:cloud,glm-5.2:cloud,glm-5.1:cloud,kimi-k2.7-code:cloud,deepseek-v4-pro:cloud,nemotron-3-super:cloud,gpt-oss:120b-cloud,qwen3-coder:480b-cloud,gemma4:cloud,m5/qwen3.6:35b-mlx"
          # cloud runs 3 at once; the Mac one at a time; both lanes parallel.
          GADFLY_PROVIDER_CONCURRENCY: "ollama-cloud=3,m5=1"
          # 3 cloud models x 3 lenses = 9 concurrent ollama-cloud queries (under the 10 budget).
          GADFLY_PROVIDER_LENS_CONCURRENCY: "ollama-cloud=3"
          # Default => the 3-lens suite (security, correctness, error-handling).
          # Set the repo var GADFLY_SPECIALISTS to override (csv / "all" / "auto").
          GADFLY_SPECIALISTS: ${{ vars.GADFLY_SPECIALISTS || 'security,correctness,error-handling' }}
          # Per-lens deadline + bounded steps so the slow local models stay sane.
          GADFLY_TIMEOUT_SECS: "600"
          GADFLY_MAX_STEPS: "14"
          # Allow-list for the comment trigger (mirrors the job-level if: guard).
          GADFLY_ALLOWED_USERS: "steve,fizi,dazed"
          # --- findings telemetry: POST runs + findings to the gadfly-reports store ---
          # Advisory & off unless GADFLY_FINDINGS_URL is set; failures only log to
          # stderr and never affect the review. GADFLY_REPO / GADFLY_PR are derived
          # in-container; the URL + token are user-scope secrets.
          GADFLY_FINDINGS_URL: ${{ secrets.GADFLY_FINDINGS_URL }}
          GADFLY_FINDINGS_TOKEN: ${{ secrets.GADFLY_FINDINGS_TOKEN }}
          # --- event context (leave as-is) ---
          EVENT_NAME: ${{ github.event_name }}
          PR: ${{ github.event.pull_request.number || github.event.issue.number || github.event.inputs.pr_number }}
          PR_BRANCH: ${{ github.head_ref }}
          IS_DRAFT: ${{ github.event.pull_request.draft }}
          COMMENT_BODY: ${{ github.event.comment.body }}
          COMMENT_ID: ${{ github.event.comment.id }}
          ACTOR: ${{ github.actor }}