* feat(kanban): goal_mode cards run workers in a /goal loop
A goal_mode card wraps its dispatched worker in the Ralph-style goal
loop behind /goal: after each turn an auxiliary judge checks the
worker's response against the card title+body, and if not done the
worker keeps going in the SAME session until the judge agrees, the
worker terminates the task itself, or the turn budget runs out (which
blocks the card for human review — never a silent exit).
- kanban_db: goal_mode + goal_max_turns columns (additive migration),
Task fields, create_task params, INSERT wiring, created-event payload.
- kanban_tools: goal_mode/goal_max_turns on the kanban_create tool so
orchestrators can opt cards in when fanning out.
- kanban CLI: --goal / --goal-max-turns on 'kanban create'.
- dashboard API: goal_mode/goal_max_turns on the create endpoint
(auto-surfaced back via asdict).
- _default_spawn: sets HERMES_KANBAN_GOAL_MODE / _GOAL_MAX_TURNS only
when the card opts in.
- goals.run_kanban_goal_loop: standalone, callback-injected loop engine
(no SessionDB persistence; ephemeral worker). cli.py quiet path calls
it after the worker's first turn when the env vars are set.
- Docs: orchestrator skill + kanban feature page.
Tests: DB roundtrip + legacy migration, spawn env gating, and the loop's
continuation/completion/budget-block/finalize-nudge branches. E2E run
against a real kanban DB confirms a budget-exhausted goal worker lands
in a sticky blocked state.
* feat(kanban/dashboard): goal-mode toggle in the create form
Wires the goal_mode card setting into the dashboard UI (the plugin's
hand-written IIFE bundle, no build step):
- InlineCreate: 'goal mode' checkbox after the skills field; checking it
reveals an optional 'max turns' number input. Both reset on submit and
only post goal_mode/goal_max_turns when enabled.
- TaskDrawer: a 'Goal mode: on (max N turns)' MetaRow so a card's
goal-mode setting is visible after creation (auto-fed by asdict via the
existing _task_dict).
Live-tested through the running dashboard with a browser: created a
goal-mode card with max-turns=8, confirmed it persisted to the kanban DB
(goal_mode=1, goal_max_turns=8) and rendered back in the drawer as
'on (max 8 turns)'. No JS console errors.
The skill enumerated 8 specialist profile names (researcher, analyst,
writer, reviewer, backend-eng, frontend-eng, ops, pm) as "the standard
roster" and told orchestrators to "assume these exist." Almost no real
Hermes setup matches that fleet — single-profile setups, Docker-worker
setups, and curated-team setups all violate it — so following the skill
literally produced cards assigned to non-existent profiles, which the
dispatcher silently failed to spawn (no autocorrect, no fallback, just
sits in `ready` forever).
Changes:
- Drop the standard-specialist-roster table.
- Add a "Profiles are user-configured — not a fixed roster" section at
the top with a Step 0 that prescribes `hermes profile list` (or asking
the user) before fanning out. Cache the result in working memory.
- Rewrite the worked task-graph example with placeholder names
(<profile-A>, <profile-B>, <profile-C>) so the structure is still
teachable but doesn't invite copy-paste of role names that may not
exist.
- Reframe the "If no specialist fits" anti-temptation rule: don't
invent profile names; ask the user.
- Add a "Inventing profile names that doesn't exist" entry to Pitfalls.
- Bump skill version 2.0.0 → 3.0.0 (semantic break: previous behavior
promised a roster the skill no longer enumerates).
- Update website/docs/user-guide/features/kanban.md to drop the
matching "(researcher, writer, analyst, backend-eng, reviewer, ops)"
line and explain the discovery prompt instead.
- Re-run website/scripts/generate-skill-docs.py to refresh the
auto-generated skill page + catalog.
Closes#21131 in spirit — addresses the same hardcoded-names footgun
@yehuosi flagged, with a different shape than their PR (delete the
roster rather than replace each name with placeholder, since the
roster table was the load-bearing footgun and the worked example is
salvageable with placeholder profile names).
Co-authored-by: yehuosi <yehuosi@users.noreply.github.com>
Completes the Windows-gating coverage for the built-in skills/ tree. Every
bundled SKILL.md now carries an explicit platforms: declaration so the
loader (agent.skill_utils.skill_matches_platform) can skip-load skills
that don't fit the current OS.
74 skills declared cross-platform (platforms: [linux, macos, windows]):
Creative (16): ascii-art, ascii-video, architecture-diagram, baoyu-comic,
baoyu-infographic, claude-design, creative-ideation, design-md,
excalidraw, humanizer, manim-video, p5js, pixel-art,
popular-web-designs, pretext, sketch, songwriting-and-ai-music,
touchdesigner-mcp
Autonomous agents: claude-code, codex, hermes-agent, opencode
Data/devops: jupyter-live-kernel, kanban-orchestrator, kanban-worker,
webhook-subscriptions, dogfood, codebase-inspection
GitHub: github-auth, github-code-review, github-issues,
github-pr-workflow, github-repo-management
Media: gif-search, heartmula, songsee, spotify, youtube-content
MCP / email / gaming / notes / smart-home: native-mcp, himalaya,
pokemon-player, obsidian, openhue
mlops (non-broken): weights-and-biases, huggingface-hub, llama-cpp,
outlines, segment-anything-model, dspy, trl-fine-tuning
Productivity: airtable, google-workspace, linear, maps, nano-pdf,
notion, ocr-and-documents, powerpoint
Red-teaming / research: godmode, arxiv, blogwatcher, llm-wiki,
polymarket
Software-dev: debugging-hermes-tui-commands, hermes-agent-skill-authoring,
node-inspect-debugger, plan, requesting-code-review, spike,
subagent-driven-development, systematic-debugging,
test-driven-development, writing-plans
Misc: yuanbao
5 skills gated from Windows (platforms: [linux, macos]):
mlops/inference/vllm (serving-llms-vllm)
vLLM is officially Linux-only; Windows requires WSL.
mlops/training/axolotl
Axolotl's flash-attn + deepspeed + bitsandbytes stack is Linux-first.
mlops/training/unsloth
Requires Triton + xformers + flash-attn — Linux only in practice.
mlops/models/audiocraft (audiocraft-audio-generation)
torchaudio ffmpeg backend + encodec dependencies are Linux-first.
mlops/inference/obliteratus
Research abliteration workflow; relies on Linux-focused pytorch
kernels and MLX — no first-class Windows path.
Same strict-over-lenient policy as the optional-skills sweep: when the
underlying tool's Windows support is rough, missing, or WSL-only, gate the
skill. Easier to un-gate after verified Windows support lands than to leak
partial support that manifests as mid-task failures.
Combined with prior commits in this branch, every bundled SKILL.md
(skills/ + optional-skills/) now has a platforms: declaration.
Workers completing a kanban task can now claim the ids of cards they
created via an optional ``created_cards`` field on ``kanban_complete``.
The kernel verifies each id exists and was created by the completing
worker's profile; any phantom id blocks the completion with a
``HallucinatedCardsError`` and records a
``completion_blocked_hallucination`` event on the task so the rejected
attempt is auditable. Successful completions also get a non-blocking
prose-scan pass over their ``summary`` + ``result`` that emits a
``suspected_hallucinated_references`` event for any ``t_<hex>``
reference that doesn't resolve.
Closes#20017.
Recovery UX (kernel + CLI + dashboard)
--------------------------------------
A structural gate alone isn't enough — operators also need to see and
act on stuck workers, especially when a profile's model is the root
cause. This PR ships the full loop:
* ``kanban_db.reclaim_task(task_id)`` — operator-driven reclaim that
releases an active worker claim immediately (unlike
``release_stale_claims`` which only acts after claim_expires has
passed). Emits a ``reclaimed`` event with ``manual: True`` payload.
* ``kanban_db.reassign_task(task_id, profile, reclaim_first=...)`` —
switch a task to a different profile, optionally reclaiming a stuck
running worker in the same call.
* ``hermes kanban reclaim <id> [--reason ...]`` and
``hermes kanban reassign <id> <profile> [--reclaim] [--reason ...]``
CLI subcommands wired through to the same helpers.
* ``POST /api/plugins/kanban/tasks/{id}/reclaim`` and
``POST /api/plugins/kanban/tasks/{id}/reassign`` endpoints on the
dashboard plugin.
Dashboard surfacing
-------------------
* ⚠ **warning badge** on cards with active hallucination events.
* **attention strip** at the top of the board listing all flagged
tasks; dismissible per session.
* **events callout** in the task drawer — hallucination events render
with a red left border, amber icon, and phantom ids as styled chips.
* **recovery section** in the task drawer with three actions: Reclaim,
Reassign (with profile picker + reclaim-first checkbox), and a
copy-to-clipboard hint for ``hermes -p <profile> model`` since
profile config lives on disk and can't be edited from the browser.
Auto-opens when the task has warnings, collapsed otherwise.
Keyed by task id so state doesn't leak between drawers.
Active-vs-stale rule: warnings clear when a clean ``completed`` or
``edited`` event supersedes the hallucination, so recovery is never
permanently stigmatising — the audit events persist for debugging but
the badge goes away once the worker succeeds.
Skill updates
-------------
* ``skills/devops/kanban-worker/SKILL.md`` documents the
``created_cards`` contract with good/bad examples.
* ``skills/devops/kanban-orchestrator/SKILL.md`` gains a "Recovering
stuck workers" section with the three actions and when to use each.
Tests
-----
* Kernel gate: verified-cards manifest, phantom rejection + audit
event, cross-worker rejection, prose scan positive + negative.
* Recovery helpers: reclaim on running task, reclaim on non-running
returns False, reassign refuses running without reclaim_first,
reassign with reclaim_first succeeds on running.
* API endpoints: warnings field present on /board and /tasks/:id,
warnings cleared after clean completion, reclaim 200 + 409 paths,
reassign 200 + 409 + reclaim_first paths.
* CLI smoke: reclaim + reassign subcommands.
Live-verified end-to-end on a dashboard with seeded scenarios:
attention strip renders, badges land on the right cards, drawer
callout shows phantom chips, Reclaim on a running task flips status to
ready + emits manual reclaimed event + refreshes the drawer,
Reassign swaps the assignee and triggers board refresh.
359/359 kanban-suite tests pass
(test_kanban_{db,cli,boards,core_functionality} + dashboard + tools).
New `hermes kanban` CLI subcommand + `/kanban` slash command + skills for
worker and orchestrator profiles. SQLite-backed task board
(~/.hermes/kanban.db) shared across all profiles on the host. Zero
changes to run_agent.py, no new core tools, no tool-schema bloat.
Motivation: delegate_task is a function call — sync fork/join, anonymous
subagent, no resumability, no human-in-the-loop. Kanban is the durable
shape needed for research triage, scheduled ops, digital twins,
engineering pipelines, and fleet work. They coexist (workers may call
delegate_task internally).
What this adds
- hermes_cli/kanban_db.py — schema, CAS claim, dependency resolution,
dispatcher, workspace resolution, worker-context builder.
- hermes_cli/kanban.py — 15-verb CLI surface and shared run_slash()
entry point used by both CLI and gateway.
- skills/devops/kanban-worker — how a profile should work a claimed task.
- skills/devops/kanban-orchestrator — "you are a dispatcher, not a
worker" template with anti-temptation rules.
- /kanban slash command wired into cli.py and gateway/run.py. Bypasses
the running-agent guard (board writes don't touch agent state), so
/kanban unblock can free a stuck worker mid-conversation.
- Design spec at docs/hermes-kanban-v1-spec.pdf — comparative analysis
vs Cline Kanban, Paperclip, NanoClaw, Gemini Enterprise; 8 patterns;
4 user stories; implementation plan; concurrency correctness.
- Docs: website/docs/user-guide/features/kanban.md, CLI reference
updated, sidebar entry added.
Architecture highlights
- Three planes: control (user + gateway), state (board + dispatcher),
execution (pool of profile processes).
- Every worker is a full OS process, spawned as `hermes -p <profile>`.
No in-process subagent swarms — solves NanoClaw's SDK-lifecycle
failure class.
- Atomic claim via SQLite CAS in a BEGIN IMMEDIATE transaction; stale
claims reclaimed 15 min after their TTL expires.
- Tenant namespacing via one nullable column — one specialist fleet
can serve many businesses with data isolation by workspace path.
Tests: 60 targeted tests (schema, CAS atomicity, dependency resolution,
dispatcher, workspace kinds, tenancy, CLI + slash surface). All pass
hermetic via scripts/run_tests.sh.
