Skip to content
BotClawAgent ObserverTracking personal agents: progress and limits
2026.10.02
KNOW WHAT CHANGED

Revisions and review

Published changes, withdrawals and source checks are separate signals. A successful fetch does not verify that a procedure still works.

Source changes awaiting review

No detected source changes are pending. This does not guarantee every source is current.

Recent published revisions

  1. r2
    · #88

    Compare OpenClaw and Hermes with the same task

    Use a controlled task and record evidence, completion time, failures and observed cost. Choose for your deployment and workflow constraints rather than a general score.

  2. r2
    · #86

    Check an upgrade against a recorded baseline

    Record the actual CLI and gateway versions, preserve recovery data, and recheck chat, tools, memory, messaging and schedules after upgrading.

  3. r2
    · #84

    Verify a backup by restoring into a separate directory

    Inventory and verify an OpenClaw archive, then restore into a test target. Keep credentials private; an archive existing on disk is not a recovery test.

  4. r2
    · #82

    Build a permission boundary that can be tested

    Separate model instructions, tool approvals and operating-system access. Use an allowed read and a denied write to test the effective boundary.

  5. r2
    · #80

    Connect one messaging channel with an explicit allowlist

    Verify sender access, conversation routing and actual reply delivery separately. Start with a private test chat; gateway registration alone is not a successful integration.

  6. r2
    · #78

    Validate one scheduled task before making it recurring

    Check timezone, execution logs, destination and cleanup with a single harmless event. The current OpenClaw docs use automations; cron remains an alias.

  7. r2
    · #76

    Inspect a skill before installing and testing it

    Review files, commands, permissions and outbound destinations before adding a skill. Keep a reversible install record and verify removal in a new session.

  8. r2
    · #74

    Check that a memory update changes the next session

    A two-session test verifies persistence, correction and retrieval with a harmless fictional preference. A verbal promise to remember does not count as a saved memory.

  9. r2
    · #72

    Verify the first Hermes conversation before connecting apps

    A small acceptance procedure separates installation, provider connectivity, tool execution and session persistence. Use harmless local input before adding messaging or schedules.

  10. r2
    · #70

    Build a repeatable RSS digest with source links

    A runnable Node example fetches feeds, removes duplicate URLs and remembers seen items. Live inputs produced 35 unique entries; an unchanged replay produced zero new entries. Agent scheduling remains untested.

  11. r1
    · #68

    openclaw 2026.7.1-1

    ### Fixes - **Codex progress replies:** keep app-server turns running after delivered progress messages so GPT/Codex reaches its authoritative terminal response instead of stopping mid-turn. (#106961, #108487) Thanks @joshavant. - **Memory

  12. r1
    · #67

    openclaw 2026.7.1-2

    ### Fixes - **npm plugin updates:** accept singleton-array metadata from newer npm clients so tracked official plugins can install and update to correction releases. (#108336)

  13. r1
    · #66

    openclaw 2026.6.34

    ## 2026.6.34 ### Highlights - **Safer browser and network boundaries:** sandboxed browser routes, trusted DNS targets, custom browser origins, and loopback provider endpoints now reject unsafe access paths. (#97958, #38290, #103075, #110693

  14. r1
    · #65

    OpenClaw 2026.8.1

    ## 2026.8.1 2026.8.1 release notes > **Update help:** If the automatic update fails, use a local coding harness to help complete the update, diagnose any migration errors, and verify that the Gateway starts correctly. Back up your configura

  15. r1
    · #64

    openclaw 2026.8.2

    ## 2026.8.2 ### Highlights - **Your Home agent, beside your work:** open Home in a right or bottom dock with `Cmd/Ctrl+Shift+H`, keep your current page in view, and preview or remove its work-context snapshot or attach selected text to your

  16. r1
    · #63

    openclaw 2026.9.1

    ## 2026.9.1 ### Highlights - **Diagrams in every chat:** Mermaid blocks now render as diagrams in the Control UI and in the native macOS, iOS, and Android apps, with enlarge previews and a retry when a diagram fails to render on mobile. (#1

  17. r1
    · #62

    openclaw 2026.9.2

    ## 2026.9.2 ### Highlights - **Faster, more responsive chat:** keep chat, dashboards, and session interactions responsive while long transcripts and disk usage are processed, with direct dashboard lookup, less cold-load work, and durable hi

  18. r1
    · #61

    openclaw 2026.9.3

    ## 2026.9.3 ### Highlights - **Safer updates:** rehearse core and plugin changes in isolated candidate state before activation, support eligible 2026.9.2 migrations, and recover abandoned update records without stopping a healthy matching G

  19. r1
    · #60

    openclaw 2026.6.35

    ## 2026.6.35 This is the final June 2026 Extended Stable (LTS) release. ### Highlights - **Safer provider and channel boundaries:** bundled providers and channel adapters now bound untrusted response bodies, reject oversized inputs before e

  20. r1
    · #59

    openclaw 2026.9.4

    ## OpenClaw v2026.9.4 20 direct commits · 1,558 pull requests · 294 contributors ### Changes included in this release The release notes and changelog contain the same content, presented in two formats: - Release notes — formatted for people

Linked documentation checks

One approved OpenClaw/Hermes page is checked per scheduled run, no more than once per seven days. The first check establishes a baseline; later page changes require editorial review.

Detection covers returned collector items and these documentation pages. Page hashes flag changes, not their meaning; no model is called. Old article changes outside a collector’s current feed window may be missed.