Skip to content

Commit 59da055

Browse files
authored
ci: add pullfrog workflow (#157)
## Summary - Add `.github/workflows/pullfrog.yml`, mirrored from varlock's copy - Includes the Azure GPT-5.6-Sol routing (`AZURE_RESOURCE_NAME`/`PULLFROG_MODEL`) and the bumpy-bot skip logic for `bumpy/version-packages` refresh runs ## Test plan - [ ] Pullfrog app installed + `AZURE_API_KEY` configured for this repo in the Pullfrog dashboard - [ ] Trigger a run (e.g. `@pullfrog` mention on a PR) and confirm it executes - [ ] Confirm a bumpy-bot-triggered run on `bumpy/version-packages` is skipped
1 parent 2e91e3a commit 59da055

1 file changed

Lines changed: 45 additions & 0 deletions

File tree

.github/workflows/pullfrog.yml

Lines changed: 45 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -0,0 +1,45 @@
1+
# PULLFROG ACTION — DO NOT EDIT EXCEPT WHERE INDICATED
2+
name: Pullfrog
3+
run-name: ${{ inputs.name || github.workflow }}
4+
on:
5+
workflow_dispatch:
6+
inputs:
7+
prompt:
8+
type: string
9+
description: Agent prompt
10+
name:
11+
type: string
12+
description: Run name
13+
14+
permissions:
15+
contents: read
16+
17+
jobs:
18+
pullfrog:
19+
# Skip runs triggered by bumpy-bot (the automated "Versioned release" PRs on
20+
# bumpy/version-packages). Every push to main refreshes that PR, which would
21+
# otherwise trigger a paid re-review each time. Human-triggered runs on that
22+
# PR (e.g. an @pullfrog mention) still execute, since the triggerer is the human.
23+
if: ${{ !contains(inputs.prompt, '"triggerer":"bumpy-bot"') }}
24+
runs-on: ubuntu-latest
25+
permissions:
26+
id-token: write
27+
contents: read
28+
steps:
29+
- name: Checkout code
30+
uses: actions/checkout@v6
31+
with:
32+
fetch-depth: 1
33+
- name: Run agent
34+
uses: pullfrog/pullfrog@v0
35+
with:
36+
prompt: ${{ inputs.prompt }}
37+
env:
38+
# Review with GPT Sol via our Azure Foundry deployment instead of the
39+
# default Anthropic route. AZURE_API_KEY comes from the Pullfrog
40+
# dashboard; these two are set here so they stay readable in run logs
41+
# (dashboard-stored values get registered as GitHub Actions log masks).
42+
# The Azure deployment must be named exactly `gpt-5.6-sol`: the
43+
# @ai-sdk/azure provider uses the model id as the deployment name.
44+
AZURE_RESOURCE_NAME: pullfrog-resource
45+
PULLFROG_MODEL: azure/gpt-5.6-sol

0 commit comments

Comments
 (0)