@kosuke-ai/kosuke 1.0.46 → 1.0.48

package/CHANGELOG.md

@@ -1,5 +1,17 @@
 # Changelog
 
+## 1.0.48
+
+- Command names from before the command grammar no longer run; `kosuke <group> --help` lists the current ones. The MCP server refuses the former toolsets too: a client that requests `deploy`, `members` or `personal-tasks` now gets HTTP 400 and must switch its configuration to `deploys`, `workspaces` or `tasks`. The `user` and `workspaces` toolsets no longer include the tools that moved to `crons`, `files`, `integrations`, `plugins`, `providers`, `sensei` and `skills`: request those toolsets instead.
+- Add `kosuke sensei threads create`, which lets Sensei open a new thread with its own Turn from any point of a Turn.
+- Add `kosuke sessions env sync` to deliver a project's current environment variables to a running session sandbox and restart its services.
+- `kosuke sessions skill-runs --help` now lists every `--result` value `finish` accepts. A marker it rejects, such as `--result pass`, still exits 0 and sends nothing, but prints one line on stderr saying it was not recorded and why.
+
+## 1.0.47
+
+- `kosuke deploys queue steer` now steers on any deploy model whose runtime takes a steer, not Claude only.
+- Add `kosuke projects batch-merging picks set` for a Batch's own session: it lists the task pull requests merged into the Batch branch, which replaces the Session Label a Batch used to add.
+
 ## 1.0.46
 
 - `kosuke sessions go-ahead set` now marks a draft pull request ready for review on GitHub, as the member, before it saves the go-ahead. When GitHub refuses, nothing is saved and the command fails with GitHub's reason.

package/README.md

@@ -40,10 +40,14 @@ kosuke sessions skill-runs finish kosuke:ship --result ready
 
 The sandbox supplies the session identity and its Turn credential. The server selects the
 running Chat Job. These commands accept only known switch names and result values, never
-free text. A child skill starts inside its parent and ends before the parent resumes.
+free text. `--result` takes one of `passed`, `failed`, `blocked`, `skipped`, `prototype`,
+`question`, `ready`, `cancelled` or `open_items`, and `--help` lists them. A child skill
+starts inside its parent and ends before the parent resumes.
 
 Markers are best effort: one request, a 1.5-second deadline, no retry, and exit status zero
-even for invalid input or a tracking failure. They print no success message. Without the
+even for invalid input or a tracking failure. They print no success message. Input they
+reject sends nothing and prints one line on stderr saying the marker was not recorded and
+why, without repeating the input. Without the
 main-agent marker context, they send no request. Delegated Runs do not record markers.
 Other commands keep their normal error exit codes. `sessions skill-runs --capabilities`
 prints `skill-markers-v1` without credentials, so an older CLI never receives an unsupported rule.

package/dist/commands/skill-runs.js

@@ -31,6 +31,13 @@ const RESULTS = new Set([
 ]);
 const UUID = /^[0-9a-f]{8}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{4}-[0-9a-f]{12}$/i;
 const DEADLINE_MS = 1500;
+const RESULT_CHOICES = [...RESULTS].join('|');
+// `run` answers --help before Commander, so this is the only help an agent reads: it names
+// every result, or an agent following it must guess one.
+const USAGE = `kosuke sessions skill-runs start <name> [--fix] [--sync] [--no-review]
+kosuke sessions skill-runs finish <name> [--result ${RESULT_CHOICES}]
+Best effort: one request, 1.5-second deadline, always exits 0.
+`;
 async function appendMarker(path, value) {
     try {
         const stat = await lstat(path).catch(() => null);
@@ -91,37 +98,49 @@ export async function runSkillMarkerCommand(args, options = {}) {
     const [group, noun] = HAND_WRITTEN.sessionsSkillRuns.command;
     if (args[0] !== group || args[1] !== noun)
         return false;
-    const { env = process.env, fetch: request = globalThis.fetch, stdout = (text) => process.stdout.write(text), timeoutMs = DEADLINE_MS, } = options;
+    const { env = process.env, fetch: request = globalThis.fetch, stdout = (text) => process.stdout.write(text), stderr = (line) => process.stderr.write(line + '\n'), timeoutMs = DEADLINE_MS, } = options;
+    // Still exit 0, but say why, so an agent can correct a rejected marker. Never echo the input:
+    // it may be free text.
+    const notRecorded = (why) => {
+        try {
+            stderr(`kosuke: skill marker not recorded: ${why}. See kosuke sessions skill-runs --help.`);
+        }
+        catch {
+            // Reporting is best effort too.
+        }
+        return true;
+    };
     if (args.length === 3 && args[2] === '--capabilities') {
         stdout('skill-markers-v1\n');
         return true;
     }
     if (args.includes('--help') || args.includes('-h')) {
-        stdout('kosuke sessions skill-runs start <name> [--fix] [--sync] [--no-review]\nkosuke sessions skill-runs finish <name> [--result <verdict>]\nBest effort: one request, 1.5-second deadline, always exits 0.\n');
+        stdout(USAGE);
         return true;
     }
     const [, , verb, rawName, ...rest] = args;
-    // Alias: `end`, the verb before the grammar, which older Factory Skills still run.
-    const action = verb === 'finish' || verb === 'end' ? 'end' : verb;
-    if ((action !== 'start' && action !== 'end') ||
-        !rawName ||
-        !/^(?:kosuke:)?[a-z][a-z0-9-]{0,79}$/.test(rawName))
-        return true;
+    if (verb !== 'start' && verb !== 'finish')
+        return notRecorded('use start or finish');
+    if (!rawName || !/^(?:kosuke:)?[a-z][a-z0-9-]{0,79}$/.test(rawName))
+        return notRecorded('give a skill name, such as kosuke:ship');
     const flags = [];
+    const action = verb === 'finish' ? 'end' : 'start';
     let result;
     for (let index = 0; index < rest.length; index++) {
         const value = rest[index];
         if (value === '--result' && action === 'end' && result === undefined) {
             result = rest[++index];
             if (!result || !RESULTS.has(result))
-                return true;
+                return notRecorded(`--result takes one of ${RESULT_CHOICES}`);
         }
         else if (FLAGS.has(value)) {
             if (!flags.includes(value))
                 flags.push(value);
         }
+        else if (value === '--result')
+            return notRecorded(action === 'start' ? 'only finish takes --result' : '--result is given once');
         else
-            return true;
+            return notRecorded('an argument is not a known switch');
     }
     const bridge = env.KOSUKE_SKILL_MARKER_FILE;
     const clockId = env.KOSUKE_SKILL_CLOCK_ID;
@@ -225,7 +244,7 @@ export function attachSkillMarkerCommands(program, options = {}) {
         for (const flag of FLAGS)
             command.option(flag);
         if (action === 'finish')
-            command.option('--result <verdict>', 'A known result, such as passed, failed, blocked, or ready.');
+            command.option('--result <verdict>', `One known result: ${RESULT_CHOICES}.`);
         command.action(async (_name, _flags, cmd) => {
             await runSkillMarkerCommand(cmd.parent.parent.parent.args, options);
         });

package/dist/generated/commands/deploys/queue-steer.js

@@ -6,7 +6,7 @@ export const operation = {
         "queue",
         "steer"
     ],
-    "summary": "Send your own queued Deploy Chat message into the running deploy agent Turn as a steer. Claude deploys only; the message stays queued when nothing can take it.",
+    "summary": "Send your own queued Deploy Chat message into the running deploy agent Turn as a steer, on any deploy model whose runtime takes one; the message stays queued when nothing can take it.",
     "pathParams": [
         "projectId",
         "queuedMessageId"