root/fusion
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

feat(FN-3754): unify agent action gating classifications and add even cards

Browse Source 
The merge introduces the new `fusion-plugin-even-cards` workspace plugin providing board card endpoints and API key authentication, while also unifying agent action gating classifications across permanent agents with configurable display limits. The GitHub integration receives review state refresh i

Fusion-Task-Id: FN-3754
This commit is contained in:
Fusion
2026-05-08 13:52:06 -07:00
committed by gsxdsm
parent d57daf13c0
commit ded66d0488
1 changed files with 52 additions and 5 deletions
Download Patch File Download Diff File Expand all files Collapse all files

57
packages/engine/src/executor.ts
View File

@@ -3916,19 +3916,18 @@ export class TaskExecutor {
};
}
const stepIndex = step - 1;
if (!Number.isInteger(stepIndex) || stepIndex < 0) {
if (!Number.isInteger(step) || step < 0) {
return {
content: [{
type: "text" as const,
text: `Invalid step number: ${step}. Steps are 1-indexed.`,
text: `Invalid step number: ${step}. Steps are 0-indexed.`,
}],
details: {},
};
}
const task = await store.updateStep(taskId, stepIndex, status as StepStatus);
const stepInfo = task.steps[stepIndex];
const task = await store.updateStep(taskId, step, status as StepStatus);
const stepInfo = task.steps[step];
const persistedStatus = stepInfo.status;
const progress = task.steps.filter((s) => s.status === "done").length;
@@ -4198,6 +4197,29 @@ export class TaskExecutor {
reviewerLog.log(`${taskId}: ${reviewType} review for Step ${step} (${step_name})`);
await store.logEntry(taskId, `${reviewType} review requested for Step ${step} (${step_name})`);
// Auto-advance the step to "in-progress" if the agent is requesting a
// review without having flipped it themselves. Some runtimes (notably
// permanent-agent CEO sessions on the openai-codex transport) skip the
// bookkeeping fn_task_update call entirely. Tying step state to the
// review tool keeps the dashboard accurate without a second tool call.
// Skip the auto-update if the step is already done/skipped (don't
// regress completed work) — updateStep guards against that anyway.
try {
const currentTask = await store.getTask(taskId);
if (
Number.isInteger(step) &&
step >= 0 &&
step < currentTask.steps.length &&
currentTask.steps[step].status === "pending"
) {
await store.updateStep(taskId, step, "in-progress");
}
} catch (autoUpdateErr) {
reviewerLog.warn(
`${taskId}: failed to auto-advance Step ${step} to in-progress on review entry: ${autoUpdateErr instanceof Error ? autoUpdateErr.message : String(autoUpdateErr)}`,
);
}
try {
const settings = await store.getSettings();
// Run the reviewer via semaphore.runNested so its slot accounting
@@ -4268,6 +4290,31 @@ export class TaskExecutor {
codeReviewVerdicts.set(step, "REVISE");
} else if (result.verdict === "APPROVE") {
codeReviewVerdicts.delete(step);
// Auto-mark the step as done once its code review passes. The
// recoverApprovedStepsOnResume path (executor.ts) already does
// this on engine restart from log scan; doing it inline avoids
// depending on the agent's follow-up fn_task_update call, which
// permanent-agent runtimes routinely skip.
try {
const currentTask = await store.getTask(taskId);
if (
Number.isInteger(step) &&
step >= 0 &&
step < currentTask.steps.length &&
currentTask.steps[step].status !== "done" &&
currentTask.steps[step].status !== "skipped"
) {
await store.updateStep(taskId, step, "done");
await store.logEntry(
taskId,
`Step ${step} (${step_name}) auto-marked done by code review APPROVE`,
);
}
} catch (autoDoneErr) {
reviewerLog.warn(
`${taskId}: failed to auto-mark Step ${step} done after APPROVE: ${autoDoneErr instanceof Error ? autoDoneErr.message : String(autoDoneErr)}`,
);
}
}
}