Summary

• Fixes Skill config requirements with object format show "[object Object]" instead of proper values #79488 by normalizing object-shaped metadata.openclaw.requires.config entries through their path field so skill status/info no longer prints [object Object] or blocks valid documented requirements.
• Fixes [Bug]: SPAWN_ALLOWLIST environment variable is ignored, preventing agent spawning #79490 by honoring OPENCLAW_SPAWN_ALLOWLIST / SPAWN_ALLOWLIST as fallback subagent target allowlists when config does not provide one.
• Fixes [Bug]: "Stop" button has no effect after the agent has reached maximum context length #79525 by mapping backend agent run ids back to originating WebChat client runs so context-overflow/error completions clear the active run and unblock Stop / queued /new lifecycle state.
• Fixes openclaw status reports gateway service as failed while systemd user unit is active #79515 by preferring systemctl is-active when systemctl show reports a stale failed state for the Gateway user service.
• Rebased onto current upstream/main so the production dependency audit includes the fast-uri pin already present on main.

Real behavior proof

• Behavior or issue addressed: Four user-facing regressions: skill config requirements rendered object metadata as [object Object], subagent spawn env allowlists were ignored, WebChat run state could stay active after backend error/context-overflow completion, and openclaw status could report a systemd Gateway service failed even when systemctl is-active reported active.
• Real environment tested: Local Windows checkout at 705723b9d0 after rebasing onto upstream/main; Node v22.15.0 for repo test wrappers; WSL Ubuntu has systemd installed but no user bus and no Linux Node, so it could not provide honest live user-systemd proof on this host.
• Exact steps or command run after this patch: Ran the touched skill/subagent/chat/frontmatter tests in one grouped repo invocation, ran the focused systemd stale-state regression, ran production dependency audit, and ran the changed gate from the rebased branch.
• Evidence after fix: Copied terminal output summary from the final local run:

OPENCLAW_VITEST_MAX_WORKERS=1 pnpm test src/agents/skills.buildworkspaceskillstatus.test.ts src/agents/openclaw-tools.subagents.sessions-spawn.allowlist.test.ts src/agents/subagent-target-policy.test.ts src/agents/tools/agents-list-tool.test.ts src/shared/frontmatter.test.ts src/gateway/server-methods/chat.directive-tags.test.ts -- --reporter=verbose
[test] passed 3 Vitest shards in 84.38s
unit-fast: 17 passed
agents: 30 passed
gateway: 74 passed

OPENCLAW_VITEST_MAX_WORKERS=1 pnpm test src/daemon/systemd.test.ts -t "prefers is-active" -- --reporter=verbose
✓ systemd runtime parsing > prefers is-active when show reports a stale failed state
Test Files 1 passed (1); Tests 1 passed | 60 skipped (61)

node scripts/pre-commit/pnpm-audit-prod.mjs --audit-level=high
No high or higher advisories found for production dependencies.

• Observed result after fix: Object-shaped skill requirements normalize correctly, SPAWN_ALLOWLIST fallback allowlists are respected, chat final/error lifecycle maps the backend run id to the client run, and the systemd parser prefers active is-active output over stale failed show state. The dependency-audit failure from the old branch base is gone after rebase.
• What was not tested: A live Linux user-systemd openclaw status run could not be executed on this Windows host because WSL reports systemctl --user bus unavailable and has no Linux Node/pnpm. The full src/daemon/systemd.test.ts file also has unrelated Windows-only failures around process.geteuid and quoted Windows EnvironmentFile paths; the new stale-state regression passed in isolation.

Verification

• pnpm docs:list
• pnpm exec oxfmt --check --threads=1 CHANGELOG.md docs/gateway/config-tools.md docs/tools/creating-skills.md docs/tools/skills.md docs/tools/subagents.md src/agents/acp-spawn.ts src/agents/openclaw-tools.subagents.sessions-spawn.allowlist.test.ts src/agents/skills.buildworkspaceskillstatus.test.ts src/agents/subagent-spawn.ts src/agents/subagent-target-policy.test.ts src/agents/subagent-target-policy.ts src/agents/tools/agents-list-tool.test.ts src/agents/tools/agents-list-tool.ts src/daemon/systemd.test.ts src/daemon/systemd.ts src/gateway/server-methods/chat.directive-tags.test.ts src/gateway/server-methods/chat.ts src/shared/frontmatter.test.ts src/shared/frontmatter.ts
• OPENCLAW_VITEST_MAX_WORKERS=1 pnpm test src/agents/skills.buildworkspaceskillstatus.test.ts src/agents/openclaw-tools.subagents.sessions-spawn.allowlist.test.ts src/agents/subagent-target-policy.test.ts src/agents/tools/agents-list-tool.test.ts src/shared/frontmatter.test.ts src/gateway/server-methods/chat.directive-tags.test.ts -- --reporter=verbose
• OPENCLAW_VITEST_MAX_WORKERS=1 pnpm test src/daemon/systemd.test.ts -t "prefers is-active" -- --reporter=verbose
• node scripts/pre-commit/pnpm-audit-prod.mjs --audit-level=high
• pnpm check:changed --base upstream/main
• git diff --check upstream/main..HEAD

Note: this shell's default Node is v22.15.0, so pnpm emitted the repo engine warning (>=22.16.0 required); the commands above completed with that warning.