Skip to content

feat(kimi): give Kimi an agentic loop via OpenCode #10

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Draft
Oppen wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

feat(kimi): give Kimi an agentic loop via OpenCode #10

Changes from all commits
Commits
Show all changes
1 commit
Select commit
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
184 changes: 72 additions & 112 deletions .github/workflows/ai-review-kimi.yml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -16,21 +16,6 @@ on:
required: false
type: string
default: ''
max_diff_lines:
description: 'Maximum lines of diff to send to API'
required: false
type: number
default: 10000
max_tokens:
description: 'Maximum tokens for AI response'
required: false
type: number
default: 4096
temperature:
description: 'AI temperature setting'
required: false
type: number
default: 1
secrets:
KIMI_API_KEY:
description: 'Moonshot AI API key'
Expand All @@ -56,9 +41,11 @@ jobs:
ref: ${{ github.event.pull_request.head.sha || format('refs/pull/{0}/head', github.event.issue.number) }}

- name: Load prompt
id: load_prompt
id: prompt
env:
INPUT_PROMPT: ${{ inputs.prompt }}
run: |
if [ -n "${{ inputs.prompt }}" ]; then
if [ -n "$INPUT_PROMPT" ]; then
cat << 'PROMPT_EOF' > /tmp/prompt.txt
${{ inputs.prompt }}
PROMPT_EOF
Expand All @@ -83,140 +70,113 @@ jobs:
PROMPT_EOF
echo "source=default" >> $GITHUB_OUTPUT
fi
echo "content<<EOF" >> $GITHUB_OUTPUT
cat /tmp/prompt.txt >> $GITHUB_OUTPUT
echo "" >> $GITHUB_OUTPUT
echo "EOF" >> $GITHUB_OUTPUT

- name: Get PR diff
id: diff
env:
GH_TOKEN: ${{ github.token }}
MAX_LINES: ${{ inputs.max_diff_lines }}
PR_NUMBER: ${{ github.event.pull_request.number || github.event.issue.number }}
run: |
gh pr diff $PR_NUMBER --repo ${{ github.repository }} > /tmp/pr_diff.txt

TOTAL_LINES=$(wc -l < /tmp/pr_diff.txt)

# Truncate if too large (Kimi has context limits)
head -n $MAX_LINES /tmp/pr_diff.txt > /tmp/pr_diff_truncated.txt
gh pr diff "$PR_NUMBER" --repo "${{ github.repository }}" > /tmp/pr_diff.txt
echo "Diff has $(wc -l < /tmp/pr_diff.txt) lines"

if [ "$TOTAL_LINES" -gt "$MAX_LINES" ]; then
echo "truncated=true" >> "$GITHUB_OUTPUT"
echo "total_lines=$TOTAL_LINES" >> "$GITHUB_OUTPUT"
echo "::warning::Diff truncated from $TOTAL_LINES to $MAX_LINES lines. Some changes were not reviewed."
else
echo "truncated=false" >> "$GITHUB_OUTPUT"
fi

- name: Call Kimi API for review
id: kimi_review
- name: Install opencode
env:
OPENCODE_VERSION: '1.15.13'
OPENCODE_SHA256: '51785a07c009f27c5125a5235c5a0ff78433dace07f73fbc44eb672ead4b3d79'
run: |
url="https://github.com/anomalyco/opencode/releases/download/v${OPENCODE_VERSION}/opencode-linux-x64.tar.gz"
curl -fsSL "$url" -o /tmp/opencode.tar.gz
echo "${OPENCODE_SHA256} /tmp/opencode.tar.gz" | sha256sum -c -
mkdir -p "$HOME/.opencode/bin"
tar -xzf /tmp/opencode.tar.gz -C "$HOME/.opencode/bin"
chmod 755 "$HOME/.opencode/bin/opencode"
echo "$HOME/.opencode/bin" >> "$GITHUB_PATH"

- name: Run Kimi Code Review
id: kimi-review
# No GITHUB_TOKEN here: the agent emits, the next step posts.
env:
KIMI_API_KEY: ${{ secrets.KIMI_API_KEY }}
MOONSHOT_API_KEY: ${{ secrets.KIMI_API_KEY }}
MODEL: ${{ inputs.model }}
MAX_TOKENS: ${{ inputs.max_tokens }}
TEMPERATURE: ${{ inputs.temperature }}
PROMPT_CONTENT: ${{ steps.prompt.outputs.content }}
REPO: ${{ github.repository }}
PR_NUMBER: ${{ github.event.pull_request.number || github.event.issue.number }}
run: |
SYSTEM_PROMPT=$(cat /tmp/prompt.txt)
echo "::add-mask::$KIMI_API_KEY"
mkdir -p .opencode
cat > opencode.json << 'OPENCODE_EOF'
{
"$schema": "https://opencode.ai/config.json",
"provider": {
"moonshotai": {
"npm": "@ai-sdk/openai-compatible",
"name": "Moonshot AI",
"options": { "baseURL": "https://api.moonshot.ai/v1" }
}
}
}
OPENCODE_EOF

if [ -z "$KIMI_API_KEY" ]; then
echo "error=true" >> $GITHUB_OUTPUT
echo "error_message=KIMI_API_KEY secret is not set" >> $GITHUB_OUTPUT
exit 0
fi
{
printf 'REPO: %s\nPR NUMBER: %s\n\n' "$REPO" "$PR_NUMBER"
printf '%s\n\n' "$PROMPT_CONTENT"
echo "The PR diff is available at /tmp/pr_diff.txt. Read it with: cat /tmp/pr_diff.txt"
echo "Output your review findings in markdown. Do NOT post a PR comment yourself - just output the review content."
} > /tmp/agent_prompt.txt

USER_PROMPT="REPO: ${{ github.repository }}
PR NUMBER: ${{ github.event.pull_request.number || github.event.issue.number }}

Review this PR diff and provide feedback on any issues found. Be specific about file names and line numbers when possible:"

jq -n \
--arg model "$MODEL" \
--arg system "$SYSTEM_PROMPT" \
--arg user "$USER_PROMPT" \
--argjson max_tokens "$MAX_TOKENS" \
--argjson temperature "$TEMPERATURE" \
--rawfile diff /tmp/pr_diff_truncated.txt \
'{
model: $model,
messages: [
{ role: "system", content: $system },
{ role: "user", content: ($user + "\n\nDiff:\n" + $diff) }
],
temperature: $temperature,
max_tokens: $max_tokens
}' > /tmp/payload.json

RESPONSE=$(curl -s -w "\n%{http_code}" -X POST "https://api.moonshot.ai/v1/chat/completions" \
-H "Content-Type: application/json" \
-H "Authorization: Bearer $KIMI_API_KEY" \
-d @/tmp/payload.json)

HTTP_CODE=$(echo "$RESPONSE" | tail -n1)
BODY=$(echo "$RESPONSE" | sed '$d')

if [ "$HTTP_CODE" != "200" ]; then
ERROR_DETAIL=$(echo "$BODY" | jq -r '.error.message // .error // .' 2>/dev/null | head -c 500)
echo "error=true" >> $GITHUB_OUTPUT
echo "error_message=Kimi API request failed with status $HTTP_CODE: $ERROR_DETAIL" >> $GITHUB_OUTPUT
exit 0
fi
set -o pipefail
opencode run --format json -m "moonshotai/${MODEL}" "$(cat /tmp/agent_prompt.txt)" > /tmp/events.jsonl

HAS_ERROR=$(echo "$BODY" | jq -e '.error.message' > /dev/null 2>&1 && echo "yes" || echo "no")
if [ "$HAS_ERROR" = "yes" ]; then
echo "error=true" >> $GITHUB_OUTPUT
echo "error_message=Kimi API returned an error response" >> $GITHUB_OUTPUT
exit 0
fi
# Each line is {type,timestamp,sessionID,...}; completed assistant text
# arrives as {"type":"text","part":{"text":"..."}} events.
jq -rs 'map(select(.type == "text") | .part.text) | join("\n")' /tmp/events.jsonl > /tmp/review.txt

REVIEW=$(echo "$BODY" | jq -r '.choices[0].message.content // "Error: Unexpected API response"')
echo "$REVIEW" > /tmp/review.txt
echo "success=true" >> $GITHUB_OUTPUT
if [ -s /tmp/review.txt ]; then
echo "success=true" >> "$GITHUB_OUTPUT"
else
echo "error=true" >> "$GITHUB_OUTPUT"
echo "error_message=opencode produced no review text" >> "$GITHUB_OUTPUT"
fi

- name: Post review comment
if: steps.kimi_review.outputs.success == 'true'
if: steps.kimi-review.outputs.success == 'true'
uses: actions/github-script@v7
env:
TRUNCATED: ${{ steps.diff.outputs.truncated }}
TOTAL_LINES: ${{ steps.diff.outputs.total_lines }}
MAX_LINES: ${{ inputs.max_diff_lines }}
PROMPT_SOURCE: ${{ steps.load_prompt.outputs.source }}
PROMPT_SOURCE: ${{ steps.prompt.outputs.source }}
MODEL: ${{ inputs.model }}
with:
github-token: ${{ github.token }}
script: |
const fs = require('fs');
const review = fs.readFileSync('/tmp/review.txt', 'utf8');
const truncated = process.env.TRUNCATED === 'true';
const totalLines = process.env.TOTAL_LINES;
const maxLines = process.env.MAX_LINES;
const promptSource = process.env.PROMPT_SOURCE;
const model = process.env.MODEL;

let truncationWarning = '';
if (truncated) {
truncationWarning = `\n> ⚠️ **Warning:** Diff was truncated from ${totalLines} to ${maxLines} lines. Some changes were not reviewed.\n`;
}

let promptNote = promptSource === 'custom' ? ' · custom prompt' : '';

const source = process.env.PROMPT_SOURCE;
const promptNote = (source === 'custom' || source === 'input') ? ' · custom prompt' : '';
if (review && review.trim() !== '') {
await github.rest.issues.createComment({
owner: context.repo.owner,
repo: context.repo.repo,
issue_number: context.issue.number,
body: `## 🤖 Kimi Code Review\n${truncationWarning}\n${review}\n\n---\n*Automated review by Kimi (Moonshot AI) · ${model}${promptNote}*`
issue_number: context.payload.pull_request?.number || context.payload.issue?.number,
body: `## 🤖 Kimi Code Review\n\n${review}\n\n---\n*Automated review by Kimi (Moonshot AI) · ${model}${promptNote}*`,
});
}

- name: Post error comment
if: steps.kimi_review.outputs.error == 'true'
if: steps.kimi-review.outputs.error == 'true'
uses: actions/github-script@v7
env:
ERROR_MESSAGE: ${{ steps.kimi_review.outputs.error_message }}
ERROR_MESSAGE: ${{ steps.kimi-review.outputs.error_message }}
with:
github-token: ${{ github.token }}
script: |
const errorMessage = process.env.ERROR_MESSAGE || 'Unknown error';
await github.rest.issues.createComment({
owner: context.repo.owner,
repo: context.repo.repo,
issue_number: context.issue.number,
body: `## 🤖 Kimi Code Review\n\n⚠️ Review failed: ${errorMessage}\n\n---\n*Automated review by Kimi (Moonshot AI)*`
issue_number: context.payload.pull_request?.number || context.payload.issue?.number,
body: `## 🤖 Kimi Code Review\n\n⚠️ Review failed: ${errorMessage}\n\n---\n*Automated review by Kimi (Moonshot AI)*`,
});