Founding member access recorded.
Checkout cancelled.

Signals · Hermes Agent profile

Hermes Agent

Every signal accepted for Hermes Agent. Each links to the run that produced it. The Hermes Agent profile carries the current evergreen state.

June 2026

  1. 2026-06-03 · Hermes Agent

    Docker dashboard insecure binding now requires explicit HERMES_DASHBOARD_INSECURE=1 opt-in

    • The dashboard no longer infers insecure mode from bind host, so operators whose Docker setups relied on that inference must add HERMES_DASHBOARD_INSECURE=1 explicitly or the dashboard will not bind insecurely.
    • Existing Docker and hosted deployments must update env configuration before upgrading to v0.15.1 to avoid a broken or unexpectedly-secured dashboard.
    • Verification path: upgrade to v0.15.1, set HERMES_DASHBOARD_INSECURE=1 only where intended, and confirm the dashboard binds as expected without falling back to host-derived inference.
  2. 2026-06-03 · Hermes Agent

    Promptware defense added against Brainworm-class prompt-injection attacks

    • Operators running the agent against untrusted content (web, repos, MCP tool output) gain a built-in defense layer they should validate against their own injection test cases rather than assume blanket coverage.
    • 19 security-tagged issues were closed in the same release, so the upgrade is the gate for these protections; staying on prior versions leaves the injection surface unmitigated.
    • Verification path: upgrade to v0.15.0 and run known Brainworm-class injection patterns to confirm the defense triggers before exposing the agent to untrusted input.
  3. 2026-06-03 · Hermes Agent

    Bitwarden Secrets Manager integration replaces per-provider API keys

    • Operators managing credentials must decide whether to migrate from per-provider API keys to centralized Bitwarden Secrets Manager, changing where secrets live and how they rotate.
    • Centralized secret management enables rotation and revocation that scattered per-provider keys did not; an operator wiring CI/automation must re-point credential sourcing.
    • Verification path: configure Bitwarden Secrets Manager on v0.15.0, confirm the agent resolves credentials from it, and test a rotation to verify the agent picks up the new secret.
  4. 2026-06-03 · Hermes Agent

    Kanban becomes a multi-agent orchestration platform with auto-decomposition, swarm topology, and worktree-per-task

    • Operators who ran Kanban as a task board must now decide whether to adopt orchestrator auto-decomposition and swarm topology, which turn a queue into a self-spawning multi-agent fleet with new operating state to supervise.
    • Per-task model overrides and worktree-per-task change the cost and isolation profile of every queued task; an operator must re-plan budget and concurrency.
    • Verification path: deploy v0.15.0, queue a decomposable task, and confirm the orchestrator spawns the expected sub-agents in isolated worktrees before trusting it with real work.

May 2026

  1. 2026-05-27 · Hermes Agent

    Hermes ships PyPI, lazy adapter install, native Windows beta

    composes with Aider , Cline , Codex , Continue

    • Builders who bounced off the prior clone-and-shell installer should re-evaluate Hermes — `pip install hermes-agent` plus lazy adapter install plus Windows beta plus Zed ACP Registry listing materially lower the floor.
  2. 2026-05-27 · Hermes Agent

    `hermes proxy`: local OpenAI-compatible endpoint backed by operator OAuth

    composes with Aider , Cline , Codex , Continue

    • Operators running `hermes proxy` on the documented loopback default (`--host 127.0.0.1`) inherit a low-risk posture; the proxy accepts client `Authorization` headers and strips them before attaching the Hermes OAuth upstream. Operators changing the bind to a non-loopback address must place their own auth in front of the port — the proxy itself does not authenticate local callers.
  3. 2026-05-27 · Hermes Agent

    Honcho identity mapping and credential-pool isolation

    composes with Aider , Cline , Codex , Continue

    • Multi-user gateway operators should upgrade past the Honcho commits (week of 2026-05-21) and the credential-pool isolation commit (2026-05-27) before running shared-thread deployments — these are quiet correctness fixes for cross-user contamination.
  4. 2026-05-27 · Hermes Agent

    Kanban corruption-hardening wave (post-v0.14.0)

    composes with Aider , Cline , Codex , Continue

    • Kanban-dependent multi-agent operators should treat the post-v0.14.0 line as the integrity-floor baseline; the corruption-hardening wave volume is the signal.
  5. 2026-05-12 · Hermes Agent

    Hermes drops mistralai from [all] extras after PyPI quarantine of 2.4.6

    • Operators who installed hermes-agent[all] on or around 2026-05-12 should verify whether mistralai==2.4.6 is present in their environment and remove it if so.
    • Operators needing Mistral Voxtral TTS must switch to explicit hermes-agent[mistral] install; it no longer ships in [all] while quarantine is active.
  6. 2026-05-12 · Hermes Agent

    Durable Kanban with hallucination gate, redaction-on-by-default, channel allowlists

    • Operators upgrading existing Hermes deployments must verify that secret redaction is now ON by default. Log pipelines that relied on unredacted output will see sanitized logs after upgrade.
    • Discord operators with role-gated access (`DISCORD_ALLOWED_ROLES`) should re-verify their role-scoping configuration: the guild-scoped fix (CVSS 8.1) may change behavior in cross-guild bot deployments.
    • Operators building multi-agent workflows on Hermes should evaluate the Kanban board's reliability primitives (heartbeat reclaim, zombie detection, hallucination gate, per-task retries) before building a custom coordination layer.
    • Operators using cron should evaluate `no_agent` mode for script-only automation that does not require LLM invocation.
  7. Persistent agent state is becoming a product surface

    • Developers need to know which goals, memory patches, recaps, sessions, and skill maintenance loops shaped a serious run.
  8. Accessibility is a frontier capability, not marketing polish

    • Everyday adoption depends on setup recovery, visible progress, voice/chat surfaces, readable UI, OAuth clarity, and fewer dead ends.
  9. Agent systems are growing control planes

    • Once agents coordinate across tasks, runtimes, gateways, and integrations, operators need liveness, cost, role, session, and recovery controls.
  10. Integrations are volatile; the operating loop has to be durable

    • Provider lists, plugin systems, transports, and model profiles will keep changing.
  11. Worker-native state is becoming a memory layer.

    • Recaps, memory patches, skill curators, and task state are moving into worker tools. Operators should use them, but should preserve an operator-owned record of what state governed each run.
    • Add worker-native state fields to adapter receipts: recap handles, memory patch ids, curator reports, skill reports, and resume state.
  12. Verification is becoming a worker capability.

    • Provider-native review, multi-agent execution, subagent evals, curator reports, and QA-like cloud fleets can catch useful issues, but their verdicts are not automatically the operator's truth.
    • Treat worker verification as evidence inputs. BitterQA or the run contract should still own the final evidence standard and settlement.
  13. Plugin, extension, and skill ecosystems are becoming the integration surface.

    • The practical power of worker CLIs increasingly depends on plugins, hooks, extensions, skills, and transport modules, not just the base model.
    • Adapter receipts should include enabled plugin/extension/skill surfaces and should distinguish worker-local skills from Bitter-owned memory.
  14. Provider-native long-horizon state is now table stakes.

← All signals