mirror of
https://github.com/Kilo-Org/kilocode.git
synced 2026-08-28 19:11:03 +08:00
d579774960
* fix(ci): pass --auto to the docs-sync kilo runs and keep full stderr logs Headless kilo run auto-rejects every permission ask it receives, and the GitHub runner has no user config granting bash — so without --auto the docs-sync bot's triage, edit and verify-fix calls were silently crippled whenever the agent reached for a non-allowlisted shell command (CI run 30306629290: 9 rejections, all 11 edit batches failed, exit 0). - Pass --auto immediately after "run" at all three call sites (triage.mjs, edit.mjs, docs-sync.yml Fix verify failures step) - runKilo now always writes the child's full stderr to docs-sync-out/kilo-stderr-<label>.log, on success as well as failure — the blindness that hid the defect - selftest asserts --auto non-vacuously (region-scoped source match + a stub invocation that records argv) and proves the stderr log is written on both the failure and the summary-writing success path * fix(cli): exit nonzero when a headless run auto-rejects or its session errors Non-interactive kilo run reported success for runs that accomplished nothing — a caller cannot distinguish success from a dead run, which is why the docs-sync bot had to stop trusting exit codes entirely. - Plain headless run (neither --auto nor --dangerously-skip-permissions) in which the CLI auto-rejected at least one permission ask now exits non-zero, even when the session still reaches idle afterwards, with the diagnostic: run ended with an auto-rejected permission; pass --auto for autonomous use. Deliberate contract change: any auto-rejected ask means the run was crippled, not successful. - A mid-stream session error now writes its diagnostic to stderr before emitting the json error event, so the cause is visible under --format json as well (the emit-first shape skipped the stderr write). - The old exit-0 contract lock-in test is replaced: its llm.fail fixture never published a consumable session.error, so it locked in a false premise. New tests cover both scenarios in both output formats; happy-path and --format json runs still exit 0 unchanged. * fix(ci): redact secret env values from persisted kilo stderr logs The full-stderr capture added for observability lands in uploaded CI artifacts, which are raw files (GitHub masks secrets in log streams only) on a public repo, and the runner env holds a long-lived KILO_API_KEY. Redact exact values of KEY|TOKEN|SECRET-named env vars (len >= 8) once at capture, so both the console tail and the artifact file are safe. Adds the changeset for the headless exit-code contract change. * fix(ci): redact secrets from persisted kilo stdout and harden redaction The docs-sync workflow uploads docs-sync-out/ as a public 14-day artifact, and kilo stdout was persisted raw there in two more places: triage-raw-*.txt and the edit-log.txt tee. Redact at capture in runKilo for stdout as already done for stderr, and pipe the verify-fix step's kilo stdout through a new line-wise redact-stream.mjs filter before tee. Also harden redactEnvSecrets: widen the name pattern to KEY|TOKEN|SECRET|CREDENTIAL|PASSWORD|ORG_ID|_PAT (KILO_ORG_ID is a repo secret), replace longer values first so a short secret that prefixes a longer one cannot leak the remainder, and document the exact-substring limitation. Selftest gains cases 2e (prefix ordering), 2f (stdout capture), 2g (stream filter) and a non-vacuous exact-line assertion in 2d. * fix(cli): correct the headless-exit changeset's json-format claim The auto-reject path adds a new error event to the --format json stream; only the existing event shapes are unchanged. Also state that the exit-1 rule covers a plain non-interactive --attach run that auto-rejects an ask. * fix(ci): document --auto security trade-off and deferred hardening Update comments in triage.mjs and edit.mjs to accurately describe the security implications of --auto (unrestricted bash for an agent steered by external PR content) and note that a scoped permission.bash map via KILO_CONFIG_CONTENT is the intended hardening, deferred until required shell patterns are stable. * fix(ci): raise docs-sync budgets so the backlog can actually drain --auto fixes the batches the bot attempted; it does not fix the ones it never started. In run 30306629290 (254 PRs collected, 51 docs-worthy), the wall-clock budgets deferred 54 PRs untriaged and 31 unedited without an attempt — 45 of the 60 pending rows on the rolling PR. Triage got 8 of 11 chunks in 35 min; edit got 4 of 11 batches in 50 min. Both are ceilings, not costs. A caught-up run needs ~2 chunks and ~1 batch and finishes in ~25 min, so raising them spends nothing on a normal day and drains the backlog on a bad one. 90/120 covers 20 chunks and 14 batches — 500 triaged and 70 edited PRs against a ~5 docs-worthy/day inflow — inside a 240-minute job timeout. Also strip ANSI CSI sequences in tailText, the shared path both triage and edit route their pending causes through. kilo renders its TUI to stderr, so every "Why" cell on the rolling PR currently reads "^[[0m→ ^[[0mRead packages/..." instead of the diagnostic. The persisted docs-sync-out/kilo-stderr-*.log stays raw as the debugging record. selftest case 2h asserts the pending reason is escape-free and still carries the diagnostic text; case 2i asserts each budget fits at least two units and the job timeout outlasts both, so a future edit cannot silently restore a budget too small to run anything. Both shown failing on the unmodified code first. * fix(ci): keep the docs-sync rebuild authoritative in the fix step The "Fix verify failures" step runs under `set -o pipefail` and the default `bash -e`. Once the CLI half of this PR ships, `kilo run` exits 1 on a mid-stream session error, which aborts the block before the rebuild runs: verify2.log is never written and `Re-verify status` reports VERIFIED=false even when the docs build fine. A transient provider error would flip every rolling docs PR to "Verification: failing". The agent's exit code was never the signal for this step — the rebuild is. Guard the pipeline with `|| echo ::warning::` so a nonzero kilo run is surfaced but the rebuild still decides the outcome. Extends selftest case 2b (which already parses this step) rather than adding a case: the guard must appear between the pipeline's tee and the rebuild, so a comment elsewhere in the block cannot satisfy it. Shown failing with the guard removed. --------- Co-authored-by: kiloconnect[bot] <240665456+kiloconnect[bot]@users.noreply.github.com>
204 lines
8.4 KiB
JavaScript
204 lines
8.4 KiB
JavaScript
// kilocode_change - new file
|
|
|
|
/**
|
|
* Runs the LLM edit pass over docs-sync-out/worthy.json in batches.
|
|
*
|
|
* Batching bounds each `kilo run` context (a replay window can yield dozens
|
|
* of docs-worthy PRs with large diffs). Each batch gets its own CLI session
|
|
* and writes its own summary file; results are merged into
|
|
* docs-sync-out/edit-summary.json. A batch that fails or is deferred by the
|
|
* wall-clock budget is recorded as action "pending" so the watermark holds
|
|
* back and the next run re-collects those PRs.
|
|
*
|
|
* Env: EDIT_MODEL (provider/model), KILO_API_KEY + KILO_ORG_ID (set by workflow; read natively by the kilo provider).
|
|
* Budgets: EDIT_BUDGET_MINUTES (default 50), EDIT_BATCH_TIMEOUT_MINUTES (default 15).
|
|
* Test hook: DOCS_SYNC_BACKOFF_MS replaces every retry wait when set.
|
|
*/
|
|
|
|
import fs from "node:fs"
|
|
import path from "node:path"
|
|
import { fileURLToPath } from "node:url"
|
|
import { backoffMsForAttempt, deadline, remainingMs, runKilo, sleepSync } from "./lib.mjs"
|
|
|
|
const BATCH_SIZE = 5
|
|
const ATTEMPTS = 3
|
|
const OUT_DIR = "docs-sync-out"
|
|
export const SUMMARY_FILE = ".docs-sync-summary.json"
|
|
|
|
const HERE = path.dirname(fileURLToPath(import.meta.url))
|
|
const basePrompt = fs.readFileSync(path.join(HERE, "edit-prompt.md"), "utf8")
|
|
const model = process.env.EDIT_MODEL
|
|
if (!model) throw new Error("EDIT_MODEL is required")
|
|
|
|
const EDIT_BUDGET_MINUTES = Number(process.env.EDIT_BUDGET_MINUTES) || 50
|
|
const EDIT_BATCH_TIMEOUT_MINUTES = Number(process.env.EDIT_BATCH_TIMEOUT_MINUTES) || 15
|
|
const BATCH_TIMEOUT_MS = EDIT_BATCH_TIMEOUT_MINUTES * 60 * 1000
|
|
|
|
const worthy = JSON.parse(fs.readFileSync(`${OUT_DIR}/worthy.json`, "utf8"))
|
|
const triage = JSON.parse(fs.readFileSync(`${OUT_DIR}/triage.json`, "utf8"))
|
|
const priority = new Map(triage.map((e) => [e.url, e]))
|
|
const ordered = [...worthy].sort((a, b) => {
|
|
const rank = { high: 0, medium: 1, low: 2 }
|
|
return (rank[priority.get(a.url)?.priority] ?? 1) - (rank[priority.get(b.url)?.priority] ?? 1)
|
|
})
|
|
|
|
/** @type {Map<string, string>} url → pending cause for failed/deferred batches */
|
|
const pendingCauses = new Map()
|
|
|
|
function formatCause(result) {
|
|
const bits = []
|
|
if (result.timedOut) bits.push("timed out")
|
|
if (result.exitCode !== null && result.exitCode !== undefined) bits.push(`exit ${result.exitCode}`)
|
|
if (result.stderrTail) bits.push(result.stderrTail.replaceAll("\n", " ").slice(0, 200))
|
|
return bits.join("; ") || "no diagnostic"
|
|
}
|
|
|
|
function editBatch(batch, index, budgetDeadline) {
|
|
const batchFile = `${OUT_DIR}/edit-batch-${index}.json`
|
|
const triageFile = `${OUT_DIR}/edit-batch-triage-${index}.json`
|
|
const summaryFile = `${OUT_DIR}/edit-summary-${index}.json`
|
|
fs.writeFileSync(batchFile, JSON.stringify(batch, null, 2))
|
|
fs.writeFileSync(
|
|
triageFile,
|
|
JSON.stringify(
|
|
batch.map((d) => priority.get(d.url)).filter(Boolean),
|
|
null,
|
|
2,
|
|
),
|
|
)
|
|
|
|
const prompt = `${basePrompt}
|
|
|
|
Batch specifics for this run: the PRs to handle are in the attached ${batchFile} (full details) and ${triageFile} (triage verdicts). Handle ONLY the PRs in these batch files. When finished, write your per-PR results in the summary JSON format described above to the file \`${summaryFile}\` (path relative to the repository root).`
|
|
|
|
let lastCause = "edit pass failed"
|
|
for (let attempt = 1; attempt <= ATTEMPTS; attempt++) {
|
|
const left = remainingMs(budgetDeadline)
|
|
if (left < BATCH_TIMEOUT_MS) {
|
|
lastCause = `edit budget exhausted before batch ${index} attempt ${attempt} (${Math.ceil(left / 1000)}s left, need ${EDIT_BATCH_TIMEOUT_MINUTES}m)`
|
|
console.warn(
|
|
`batch ${index}: stopping retries — remaining budget cannot fit another ${EDIT_BATCH_TIMEOUT_MINUTES}m attempt`,
|
|
)
|
|
break
|
|
}
|
|
|
|
// Headless `kilo run` auto-rejects every permission ask; without --auto the
|
|
// agent cannot run shell commands. SECURITY: --auto grants unrestricted bash
|
|
// to an agent steered by external PR content. Hardening deferred: a scoped
|
|
// permission.bash map via KILO_CONFIG_CONTENT should replace --auto once the
|
|
// required shell patterns are stable (see PR #12605 review thread).
|
|
const result = runKilo({
|
|
args: ["run", "--auto", prompt, "-m", model, "--variant", "high", "--dir", process.cwd(), "-f", batchFile, "-f", triageFile],
|
|
timeoutMs: Math.min(BATCH_TIMEOUT_MS, left),
|
|
streamStdout: true,
|
|
label: `edit batch ${index} attempt ${attempt}`,
|
|
})
|
|
|
|
if (fs.existsSync(summaryFile)) return true
|
|
// Tolerate the agent dropping the docs-sync-out/ prefix.
|
|
const alt = path.basename(summaryFile)
|
|
if (fs.existsSync(alt)) {
|
|
fs.renameSync(alt, summaryFile)
|
|
return true
|
|
}
|
|
|
|
// Exit 0 is not success: missing summary is a failure logged WITH the
|
|
// captured stderrTail and exit code on every attempt.
|
|
const cause = formatCause(result)
|
|
lastCause = `edit batch ${index}: ${cause}`
|
|
console.warn(
|
|
`batch ${index} attempt ${attempt}: summary file ${summaryFile} not produced` +
|
|
` (exit ${result.exitCode}${result.timedOut ? ", timed out" : ""})` +
|
|
(result.stderrTail ? `\nstderr tail:\n${result.stderrTail}` : "\nstderr tail: (empty)"),
|
|
)
|
|
|
|
if (attempt < ATTEMPTS) {
|
|
const wait = backoffMsForAttempt(attempt)
|
|
// Skip the wait when the remaining budget cannot fit another attempt.
|
|
const afterWait = remainingMs(budgetDeadline) - wait
|
|
if (wait > 0 && afterWait >= BATCH_TIMEOUT_MS) {
|
|
console.warn(`batch ${index}: backing off ${wait / 1000}s before attempt ${attempt + 1}`)
|
|
sleepSync(wait)
|
|
} else if (wait > 0) {
|
|
console.warn(
|
|
`batch ${index}: skipping backoff — remaining budget cannot fit attempt ${attempt + 1} after wait`,
|
|
)
|
|
}
|
|
}
|
|
}
|
|
|
|
console.warn(`::warning::edit batch ${index} failed after up to ${ATTEMPTS} attempts; ${batch.length} PRs pending`)
|
|
for (const d of batch) pendingCauses.set(d.url, lastCause)
|
|
return false
|
|
}
|
|
|
|
const batches = []
|
|
for (let i = 0; i < ordered.length; i += BATCH_SIZE) {
|
|
batches.push(ordered.slice(i, i + BATCH_SIZE))
|
|
}
|
|
console.log(`editing docs for ${ordered.length} PRs in ${batches.length} batches of up to ${BATCH_SIZE}`)
|
|
|
|
const budgetDeadline = deadline(EDIT_BUDGET_MINUTES)
|
|
let deferredFrom = -1
|
|
|
|
for (let i = 0; i < batches.length; i++) {
|
|
const left = remainingMs(budgetDeadline)
|
|
if (left < BATCH_TIMEOUT_MS) {
|
|
deferredFrom = i
|
|
const deferredPrs = batches.slice(i).reduce((n, b) => n + b.length, 0)
|
|
console.warn(
|
|
`stopping edit pass before batch ${i}: remaining budget (${Math.ceil(left / 1000)}s) cannot fit a ${EDIT_BATCH_TIMEOUT_MINUTES}m batch; deferring ${deferredPrs} PRs`,
|
|
)
|
|
const cause = `edit budget exhausted before batch ${i} (${Math.ceil(left / 1000)}s left)`
|
|
for (let j = i; j < batches.length; j++) {
|
|
for (const d of batches[j]) pendingCauses.set(d.url, cause)
|
|
}
|
|
break
|
|
}
|
|
editBatch(batches[i], i, budgetDeadline)
|
|
}
|
|
|
|
if (deferredFrom >= 0) {
|
|
console.warn(
|
|
`edit pass deferred ${batches.slice(deferredFrom).reduce((n, b) => n + b.length, 0)} PRs due to wall-clock budget`,
|
|
)
|
|
}
|
|
|
|
// Merge batch summaries. Coverage: every worthy PR gets an entry so the PR
|
|
// body accounts for it; failed/deferred batches show up as pending (not skipped).
|
|
const merged = []
|
|
const seen = new Set()
|
|
for (let i = 0; i < batches.length; i++) {
|
|
const file = `${OUT_DIR}/edit-summary-${i}.json`
|
|
let entries = []
|
|
try {
|
|
entries = JSON.parse(fs.readFileSync(file, "utf8"))
|
|
} catch {
|
|
continue
|
|
}
|
|
for (const e of entries) {
|
|
const url = String(e?.url ?? "")
|
|
if (!url.startsWith("http") || seen.has(url)) continue
|
|
seen.add(url)
|
|
merged.push({
|
|
pr: Number(e.pr) || 0,
|
|
url,
|
|
action: String(e.action ?? "skipped"),
|
|
reason: String(e.reason ?? ""),
|
|
})
|
|
}
|
|
}
|
|
for (const d of ordered) {
|
|
if (seen.has(d.url)) continue
|
|
const cause = pendingCauses.get(d.url) || "edit pass failed or timed out for this PR"
|
|
merged.push({ pr: d.number, url: d.url, action: "pending", reason: cause })
|
|
}
|
|
|
|
// upsert-pr.mjs consumes the merged summary from the repo root; the file is
|
|
// removed there before committing so it never lands in the docs PR.
|
|
fs.writeFileSync(SUMMARY_FILE, JSON.stringify(merged, null, 2))
|
|
const changed = merged.filter((e) => e.action !== "skipped" && e.action !== "pending").length
|
|
const skipped = merged.filter((e) => e.action === "skipped").length
|
|
const pending = merged.filter((e) => e.action === "pending").length
|
|
console.log(`edit pass complete: ${changed} changed, ${skipped} skipped, ${pending} pending`)
|