Adds a background-shell plugin providing the `bgsh` helper: named tmux
sessions that survive across tool calls and SSH disconnects, for work
ordinary tool calls cannot hold — interactive prompts, REPLs, TUIs, and
jobs that must outlive a connection.
The skill description is written as a hard gate: over-triggering is the
primary failure mode, so it redirects to plain Bash for everyday work.
- tty is the default mode (output stays in the pane, cd persists);
--log is the alternate for one-shot captured payloads
- a local registry keeps remote sessions findable, storing *where to
look* and never *what is running* — every read verifies against the
host, and prune refuses to act on an unreachable one (UNSURE != STALE)
- bin/bgsh is on the plugin PATH, so no symlink step is needed
- SKILL.md carries a "why bgsh does it this way" table: each row is a
real failure whose output was indistinguishable from success
Passes `claude plugin validate` with no warnings; bgsh is shellcheck
clean and bash 3.2 compatible.
Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Two shifts in scope:
- Remove the line-editor agent. The plugin no longer mutates the
manuscript — all commands report findings to the conversation
and leave revisions to the writer.
- Consolidate the consistency surface. /check-voice and
/check-continuity were narrow checks with overlapping concerns;
they are replaced by /check-consistency, a single broader pass
covering author voice, character voice, words-vs-actions
alignment, and in-world believability. Plot-logic and
sequencing concerns formerly in /check-continuity are now
served by /generate-timeline's consistency-issues section.
- Add /proofread for the mechanical layer that previously had no
command equivalent — grammar, tense, punctuation, spelling,
agreement, sentence-level clarity. Reports issues; does not edit.
Command count unchanged at nine. The plugin is now agent-free and
mutation-free; every command is read-only on the manuscript.
BREAKING CHANGE: /check-voice and /check-continuity are removed
(superseded by /check-consistency). The line-editor agent is
removed (use /proofread for reporting; the writer makes the edits).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Strip the Universalis-specific vault layout and card-management
machinery. The plugin is now a general-purpose fiction toolkit:
nine commands that take a manuscript path/glob/selection via
\$ARGUMENTS and write their output to the conversation.
Surviving commands, flattened into commands/:
- analyze-story: structural and statistical analysis
- bechdel-test: representation tests (Bechdel + related)
- check-continuity: plot/character/world continuity
- check-voice: per-character voice consistency
- developmental-review: structural feedback on a scene/chapter
- extract-research: world-building and research inventory
- final-review: synthesis of prior-phase reviews
- generate-timeline: chronology with consistency checks
- strategic-review: craft-based critique with priority ranking
Removed (vault and card management): 10 update-*-cards commands,
cleanup-generated, update-changed, update-links,
update-chapter-summaries, validate-story, validate-pdf, score-scenes.
The original bespoke command set is preserved in 65c29d7 for recovery.
Agent (line-editor) and skill (reviewing-fiction) were already
portable and are unchanged. plugin.json no longer needs the
commands subdirectory paths since the layout is flat. Marketplace
description updated to reflect the general-purpose scope.
BREAKING CHANGE: All update-*-cards, cleanup-generated, update-changed,
update-links, update-chapter-summaries, validate-story, validate-pdf,
and score-scenes commands are removed. perform-bechdel-tests is renamed
to bechdel-test. review-final is renamed to final-review.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Snapshot of the fiction-assistance plugin in its original Universalis-
specific form before generalization. Adds the missing plugin.json
manifest (subdirectory command paths explicitly listed), moves the two
stray root-level review files into their proper homes (review-all.md
into the reviewing-fiction skill, review-final.md as a command with
frontmatter), and registers the plugin in the marketplace.
Preserved here so the original card-generation and vault-maintenance
commands remain recoverable when the plugin is reduced to a portable
guidance-and-analysis subset in the next commit.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Nine new agents for fiction work: developmental editor, workshop
instructor, social media advisor, literary agent, bookseller, a
speculative novelist (Le Guin / Banks / Reynolds composite), and
three reader voices — traditional sci-fi, literary sci-fi, and an
omnivorous list-topping reader. Bumps plugin to 1.2.0 and updates
the marketplace blurb to reflect the broader roster.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Validated against the plugin-dev:skill-development checklist and
applied conformance fixes:
Description style — all five skill descriptions now use the canonical
"This skill should be used when..." third-person form, with expanded
trigger-phrase coverage. Previously used imperative "Use when..." which
the spec calls out as wrong.
Writing voice — body prose changed from "You are the dispatcher / You
walk a user through..." to imperative "Act as the dispatcher / Walk
the user through...". Remaining "you" usage is confined to quoted
output text the agent prints to the user, which is correct usage.
Progressive disclosure — composer/skills/ui-system/SKILL.md was 2513
words (over the 2000-word ideal). Refactored to 1063 words by moving
detailed phase content into:
- references/design-interview.md (D1–D11 phase specifications)
- references/interaction-interview.md (I1–I11 phase specifications)
- references/inspiration-presets.md (Linear/Notion/Stripe/etc. table)
The SKILL.md now points at each reference as needed; agents load only
what's relevant for the current phase. Other four skills already
within the 1000–2000 word range, no refactor needed.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Per plugin-structure spec, plugin-internal references in component
files (commands, agents, skills) must use \${CLAUDE_PLUGIN_ROOT} so
they resolve correctly regardless of where the plugin is installed
(local repo, marketplace cache, npm-installed, etc.).
Updated:
- symphony/skills/symphony-init/SKILL.md: WORKFLOW.md template ref
- composer/skills/ui-system/SKILL.md: all six template refs in the
Write phase
- composer/skills/composer/SKILL.md: added a Templates section that
enumerates each phase's template path with \${CLAUDE_PLUGIN_ROOT},
plus the ADR template ref in phase 8
Composer's CLAUDE.md gains a clarifying note that templates/design-
system/SKILL.md and templates/interaction-design/SKILL.md are template
skeletons, not actual plugin skills — they're copied into a consuming
project's .claude/skills/ at install-time of the design system. Per
the auto-discovery rule (scans skills/ only), they are not loaded as
plugin skills.
Removed empty composer/agents/ directory (composer has no agents;
workers are owned by symphony).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Both plugins now have .claude-plugin/plugin.json manifests and are
listed in the root marketplace.json so they can be installed via the
plugin marketplace UI alongside council-of-experts and genealogy.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Earlier framing assumed each /schedule-fired tick was a cold start
needing on-disk reconciliation. The actual model is a long-lived
orchestrator session where /schedule fires ticks within the running
session — TaskList sees prior-tick tasks just fine. Reframed
accordingly, while leaving the door open to other trigger sources
(Discord channel, webhook, second user session) which DO need
cross-session reconciliation.
Reconciliation logic now has a fast path and a slow path:
- Fast path: task in our TaskList → it's ours → reconcile via TaskGet.
- Slow path: task NOT in our TaskList (another session, or our session
restarted) → check the worker's heartbeat file. Fresh → another
session owns it, leave alone. Stale or missing → consider abandoned,
retry; do not call TaskStop on a task we don't own.
Workers now write .symphony/heartbeats/<issue-id>.json on start,
refresh it at heartbeat_interval_ms cadence (default 60s), and delete
it on graceful exit. Stale heartbeats are how abandonment is detected
across sessions and after crashes.
Also added session_id to state.json running entries so cross-session
reconcilers know who spawned what.
WORKFLOW.md gains heartbeat_interval_ms and heartbeat_stale_ms knobs
under agent: with documented defaults.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
ui-system skill is no longer a placeholder. Full design-system and
interaction-design interviews with branching, defaults, batched
question clusters, and an 'accept default / skip' protocol so users
move through fast when they have taste and patient when they don't.
Design system interview covers: visual posture, color (with WCAG
contrast verification before writing), typography, spacing, radii,
shadows, density, visual motion, components, theming, and
accessibility tokens (focus ring, hit targets).
Interaction design interview covers: input model, state-surface
declarations per action class, motion posture and triggers,
breakpoints, i18n, four state postures (empty/loading/error/offline),
time formatting, keyboard policy, optimistic-update strategy with
rollback and conflict resolution, gestures, focus management,
notification patterns, selection, and drag-and-drop.
Reference banks for "make it look like Linear/Notion/Stripe/etc."
translate well-known products into starting tokens — user still
confirms each value, no silent invention.
Templates expanded to hold the full output:
- design-system/SKILL.md grew sections for accessibility tokens,
density, dark-mode mapping, contrast verification, theming variants
- design-system/tokens.ts now exports color (light + dark), typography
(family/size/weight/line-height/letter-spacing), spacing + density-
aware padding tokens, radii, shadows, motion (durations + easings),
z-index scale, breakpoints, hit-target minimums
- design-system/components.md gained anatomy/state/a11y coverage for
10 primitives plus an "adding a new component" workflow and a
"patterns to avoid" list (nested cards, color-only meaning, etc.)
- interaction-design/SKILL.md gained sections for optimistic updates,
conflict resolution, gestures, focus management, notification
patterns, selection/multi-select, and drag-and-drop
Hard rules: never invent on the user's behalf; <TBD> instead. Verify
contrast before writing color tokens. Always declare prefers-reduced-
motion as honored.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Plugin pairs with Symphony — Composer's output (issues in the project's
tracker) is Symphony's input. State-aware /composer command walks the
user through 11 phases: concept, users (with workflow probing for hidden
actors), requirements/NFR/IA, design system, interaction design, flows,
mocks, architecture, stories, roadmap, load. Each phase produces a
reviewable artifact in docs/planning/ (or, for design and interaction,
a project-local skill — see below).
Architecture is intentionally LAST, after the spec is complete. Mocks
are JSX/TSX regardless of final stack and pass a "text feasibility"
audit — every visible string is categorized (static/computed/authored),
computed strings are traced to source data, and any string implying an
unspecified feature halts the commit.
Design system and interaction design are produced as PROJECT-LOCAL
SKILLS in .claude/skills/<slug>-design-system/ and
.claude/skills/<slug>-interaction-design/, not as docs in
docs/planning/. Skills auto-load via description matching whenever UI
work happens; a reference doc gets read once and drifts out of context.
The mocks skill imports tokens directly from the design-system skill
for a single source of truth across mocks and (eventually) real code.
ui-system is currently a PLACEHOLDER — captures the minimum needed for
mocks to proceed with <TBD: ...> markers as a return-pass worklist.
Full interview specification is the next pass.
Council touchpoints baked in: PM/skeptic/end-user after requirements,
end-user/a11y after mocks, full architecture council after architecture,
plus auto-additions (legal-triage on regulated data, privacy-advocate
on PII, finance-controller on money flows).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Sketch of a Claude Code port of openai/symphony — a daemon-style
orchestrator that polls an issue tracker, dispatches isolated worker
agents per issue, and reconciles state across ticks. Tracker-agnostic
by design: the project's CLAUDE.md documents how to talk to whatever
tracker is in use (GitHub Issues, Gitea, tracker CLI, Linear, etc.);
the plugin reads those instructions rather than shipping per-tracker
adapters.
Includes:
- /symphony-init for first-time setup (detects gh/gitea/tracker, writes
the tracker section into project CLAUDE.md, drops WORKFLOW.md, offers
to schedule the tick)
- /symphony-tick fired by /schedule cron entries (idempotent, silent,
never prompts)
- per-issue worker agent in isolated worktrees
- on-disk state in .symphony/state.json (survives cron cold starts)
Known divergences from the spec are documented in the plugin CLAUDE.md
(no streaming agent telemetry; no codex.* knobs; hooks run inside the
worker rather than the orchestrator).
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Each expert now has a personal name, background, and motive paragraph —
councils produce real disagreement instead of committee mush. Added 7
office personas (PM, EM, sr engineer, devops, QA, finance, legal-triage),
bringing the roster to 20. Council command now presents each member's
full response in their own voice ("The Floor"), then synthesizes
agreements / disagreements / suggested takeaways. add-expert template
updated to require the same shape going forward.
Co-Authored-By: Claude Opus 4.7 (1M context) <noreply@anthropic.com>
Adds new skill to quickly generate project-specific .mcp.json files with
pre-configured templates for common use cases:
- Technical: Full MCP suite for software development
- Prose: Writing-focused MCPs only
- Minimal: Just memory essentials
- Custom: User-specified MCPs
Eliminates need to look up MCP configurations for each new project.
Bump version to 1.3.0.
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
