Credits
Muster's design was inspired by atomic-claude (site), superpowers, and gsd-core. It vendors a curated set of MIT-licensed skills and agents, and every skill, agent, command, output style and pipeline that is based on an external repository or work (an article, book, paper, standard, talk or person) says which one in its own upstream: field.
How provenance is recorded
vendor/manifest.yaml is the single source of truth. Each item's upstream: field, the catalog provenance summary, and NOTICE (with each upstream's copyright line) are generated from it by muster provenance --write, and the test suite fails when any of them drift. An item with upstream: [] is Muster-authored. A repository entry names the repository, the pinned commit, the upstream path, the license, and the relation; a work entry names its kind, title, author and link. An origin Muster's history mentions but never names is recorded as unknown with a dated note saying what was searched. Relations:
| Relation | Meaning |
|---|---|
| vendored | upstream text carried essentially unchanged |
| adapted | upstream text edited |
| mechanism-port | a mechanism re-implemented in new text |
| inspired-by | an idea, structure, or framework named; no text carried over |
A work is always inspired-by: Muster takes ideas, frameworks or structure from it, never its text.
The Codex plugin carries the same record as NOTICE and PROVENANCE.md, since Codex skill headers cannot hold the field.
Vendored sources
| Source | License | Provides |
|---|---|---|
| obra/superpowers | MIT | Brainstorming, planning, TDD, code-review, debugging, verification skills |
| wshobson/agents | MIT | Software and knowledge-work agents and skills across many specialties |
| open-gsd/gsd-core | MIT | Plan, execute, and verify workflow phases (and the Codex muster-gsd-* fallbacks) |
Other sources
Every non-vendored source, generated from the manifest. Each row lists the items it is the basis for.
Repositories
| Source | License | Basis for | Relation |
|---|---|---|---|
pbakaus/impeccable @ b856e5010f2d | Apache-2.0 | 26 vendored builtins and agents | inspired-by, adapted |
obra/superpowers @ 22f0b463e7a5 | MIT | 31 vendored builtins and agents | adapted, vendored |
wshobson/agents @ 9c61ab243def | MIT | 98 vendored builtins and agents | adapted, vendored |
anthropics/skills @ 2c5958ecac7c | Apache-2.0 | 4 vendored builtins and agents | adapted, vendored |
Leonxlnx/taste-skill-classics @ 5e08c9383dc4 | MIT | 4 vendored builtins and agents | vendored |
addyosmani/agent-skills @ 301b02b12bff | MIT | 3 vendored builtins and agents | adapted, vendored |
emilkowalski/skills @ 715c1807e08c | MIT | 23 vendored builtins and agents | vendored |
nutlope/hallmark @ 7e213cb6feb4 | MIT | 109 vendored builtins and agents | adapted, vendored |
greensock/gsap-skills @ aed9cfd32777 | MIT | 8 vendored builtins and agents | vendored |
elayadesign/ai-design-skills @ 456d9e0d3e2d | MIT | 2 vendored builtins and agents | vendored |
jakubkrehel/skills @ 8b254eb75dad | MIT | 46 vendored builtins and agents | vendored |
gotalab/cc-sdd @ e2a0c671aef3 | MIT | 2 vendored builtins and agents | vendored |
K-Dense-AI/scientific-agent-skills @ 2d2065fb4520 | MIT | 36 vendored builtins and agents | vendored, adapted |
AminBlg/SimpleEnglish @ 300432a9e4c4 | MIT | 5 vendored builtins and agents | vendored |
supabase/agent-skills @ c9be0e931b79 | MIT | 33 vendored builtins and agents | vendored |
samber/cc-skills-golang @ d68d80aec298 | MIT | 28 vendored builtins and agents | vendored |
Nanako0129/sepia @ 0eb11d6c64c9 | MIT | 20 vendored builtins and agents | vendored |
haowjy/creative-writing-skills @ 15003a6fd5a1 | Apache-2.0 | 31 vendored builtins and agents | vendored |
haowjy/creative-writing-skills @ 15003a6fd5a1 | Apache-2.0 | book-developmental-editor, book-line-editor, book-continuity-checker, book-beta-reader | inspired-by |
danjdewhurst/story-skills @ 3c0e0346ee09 | MIT | book-continuity-checker, book-beta-reader, book-run | inspired-by |
Calliope-Editor/writing-skills @ 2fad031982cb | MIT | book-continuity-checker, book-run | inspired-by |
epicsagas/Velith @ 5ddd29675c5f | Apache-2.0 | book-beta-reader, book-run | inspired-by |
clutcher/book-writing-skills @ 0bea387d1fe1 | MIT | book-run | inspired-by |
PenglongHuang/chinese-novelist-skill @ 507d41567925 | MIT | book-writer | inspired-by |
howells/fiction @ 2f256acbfce6 | NONE | book-continuity-checker | inspired-by |
felipelobomotta-blip/book-genesis-studio @ 88e2e0047ad2 | MIT | book-beta-reader | inspired-by |
open-gsd/gsd-core @ 9385f4618b52 | MIT | 42 vendored builtins and agents | adapted, vendored |
nextlevelbuilder/ui-ux-pro-max-skill @ 4f2dbb67e649 | MIT | 144 vendored builtins and agents | vendored, inspired-by, adapted |
ghaida/intent @ 1806110106a4 | CC0-1.0 | 44 vendored builtins and agents | vendored, inspired-by, adapted |
bergside/typeui @ efc5f21dff38 | MIT | 4 vendored builtins and agents | adapted |
skovalik/perception-first-design @ 1704ff61c188 | CC-BY-SA-4.0 | 40 vendored builtins and agents | vendored, adapted |
ai-evals-course/evals-skills @ 80d5f7b0127c | Apache-2.0 | 7 vendored builtins and agents | vendored, inspired-by, adapted |
getsentry/skills @ e94f22214528 | Apache-2.0 | 12 vendored builtins and agents | vendored, inspired-by, adapted |
getsentry/skills @ e94f22214528 | CC-BY-SA-4.0 | 23 vendored builtins and agents | inspired-by, adapted, vendored |
openai/skills @ 49f948faa925 | Apache-2.0 | 5 vendored builtins and agents | inspired-by, adapted, vendored |
trailofbits/skills @ e4dd0bb5b155 | CC-BY-SA-4.0 | 26 vendored builtins and agents | vendored, inspired-by, adapted |
anthropics/claude-plugins-official @ dceb15d50832 | Apache-2.0 | 4 vendored builtins and agents | vendored, adapted |
mattpocock/skills @ 4ae2317ddb22 | MIT | 15 vendored builtins and agents | vendored, inspired-by, adapted |
obra/superpowers-lab @ 51111f74f240 | MIT | 8 vendored builtins and agents | vendored, inspired-by, adapted |
addyosmani/agent-skills @ 301b02b12bff | MIT | 1 vendored builtins and agents | adapted |
open-gsd/gsd-core @ 1c352d1ea37b | MIT | muster-gsd-plan-phase, muster-gsd-execute-phase, muster-gsd-verify-work | mechanism-port |
wshobson/agents @ 2de74ac1c8f6 | MIT | wsh-signed-audit-trails-recipe, wsh-code-reviewer | mechanism-port, vendored |
blader/humanizer @ b48c832f0f80 | MIT | muster-humanizer | adapted |
rudra496/StealthHumanizer @ 8b0e90874c61 | MIT | muster-humanizer | adapted |
conorbronsdon/avoid-ai-writing @ 30971f1c50c6 | MIT | muster-humanizer | adapted |
jalaalrd/anti-ai-slop-writing @ 63255f9bbb75 | NONE | muster-humanizer | inspired-by |
aaaronmiller/humanize-writing @ c07a2a8dab1f | MIT | muster-humanizer | adapted |
linexjlin/GPTs @ 3adfb7b38423 | NONE | muster-humanizer | inspired-by |
hardikpandya/stop-slop @ 8da1f030185b | MIT | muster-humanizer | adapted |
petergyang/no-ai-slop @ 000650b15698 | MIT | muster-humanizer | adapted |
hermes-labs-ai/lintlang @ 645f57879ff5 | Apache-2.0 | muster-prompt-smith | mechanism-port |
promptfoo/promptfoo @ 8569e72ed6cf | MIT | muster-prompt-smith | inspired-by |
shobrook/promptimal @ e03cbc1db19c | MIT | muster-prompt-smith | inspired-by |
gepa-ai/gepa @ ce5b0d6e3d6a | MIT | muster-prompt-smith | inspired-by |
felipelobomotta-blip/book-genesis-studio @ c2fc628fd5d1 | MIT | book, muster-scorer | mechanism-port, inspired-by |
NousResearch/autonovel @ d165f267a0ff | NONE | book, book-run | inspired-by |
CSHaitao/Awesome-LLMs-as-Judges @ d89ed4241b81 | NONE | muster-scorer | inspired-by |
llm-as-a-judge/Awesome-LLM-as-a-judge @ ad3b4ce1d869 | NONE | muster-scorer | inspired-by |
microsoft/LLM-Rubric @ 030ab161e5af | MIT | muster-scorer | inspired-by |
assafelovic/gpt-researcher @ 597a99e2ef87 | Apache-2.0 | muster-research | inspired-by |
mbzuai-nlp/qraft @ 3bbea66c832d | NONE | muster-research | inspired-by |
damusix/atomic-claude @ b4ca195cae60 | MIT | muster-surgeon, muster-reviewer, muster-investigator, muster-strategist, muster-improver, improve, muster | inspired-by |
obra/superpowers @ d884ae04edeb | MIT | muster-runner | mechanism-port |
wshobson/agents @ 511f8345cf42 | MIT | muster-runner | mechanism-port |
VoltAgent/awesome-claude-code-subagents @ 14bbcf5d984c | MIT | muster-runner | mechanism-port |
humanlayer/12-factor-agents @ d20c728368bf | Apache-2.0 AND CC-BY-SA-4.0 | muster-runner | mechanism-port |
contains-studio/agents @ a5a480c324ca | NONE | muster-runner | inspired-by |
disler/infinite-agentic-loop @ 6e9a012f81ef | NONE | muster-runner | inspired-by |
hesreallyhim/awesome-claude-code @ 957922fca4cc | CC-BY-NC-ND-4.0 | muster-runner | inspired-by |
deanpeters/product-manager-prompts @ 418049a64858 | CC-BY-NC-SA-4.0 | prd, epic | inspired-by |
deanpeters/Product-Manager-Skills @ 70fb6c4e4163 | CC-BY-NC-SA-4.0 | prd, epic, launch-plan | inspired-by |
phuryn/pm-skills @ 9af7bfd0bc47 | MIT | 8 vendored builtins and agents | inspired-by, adapted |
github/spec-kit @ 08252af8c5fb | MIT | muster-pm | inspired-by |
Fission-AI/OpenSpec @ 579fcac57ae0 | MIT | muster-pm | inspired-by |
Wirasm/prp @ 435292597e7b | MIT | muster-pm | inspired-by |
product-on-purpose/pm-skills @ 9b920db418c5 | Apache-2.0 | 6 vendored builtins and agents | inspired-by, adapted |
phuryn/product-engineering @ 061014ffe954 | MIT | 3 vendored builtins and agents | adapted |
open-gsd/gsd-core @ 9385f4618b52 | MIT | muster-pm | inspired-by |
anthropics/knowledge-work-plugins @ e605d8ddbfb6 | Apache-2.0 | 6 vendored builtins and agents | inspired-by, adapted |
joelparkerhenderson/objectives-and-key-results @ f7269521672a | NONE | okrs | inspired-by |
architecture-decision-record/architecture-decision-record @ a36c53f49517 | CC-BY-NC-SA-4.0 | design | inspired-by |
adr/madr @ ba75bb1b20d4 | MIT OR CC0-1.0 | design | inspired-by |
olivierlacan/keep-a-changelog @ 08d0df5a7e93 | MIT | release-notes | inspired-by |
gcamilo/management-consulting @ dd2d099dc390 | MIT | executive-summary | inspired-by |
reasoncommons/issuetrees @ 56ad78e0ce78 | NONE | executive-summary | inspired-by |
cdeust/ai-prd-generator @ 4189d41b102d | LicenseRef-proprietary | prd | inspired-by |
coreyhaines31/marketingskills @ 8a3bde8083bd | MIT | 8 vendored builtins and agents | inspired-by |
CosmoBlk/email-marketing-bible @ 962648c1b147 | MIT | competitive-battlecard, lead-magnet, case-study, newsletter | inspired-by |
AgriciDaniel/claude-email @ 182270a0bbcd | MIT | competitive-battlecard, lead-magnet, case-study, newsletter | inspired-by |
SkeltonThatcher/run-book-template @ 7014714b6021 | NONE | runbook | inspired-by |
dastergon/awesome-sre @ e91ef69aa521 | CC0-1.0 | runbook | inspired-by |
runbear-io/awesome-runbook @ d83a294e4eb6 | CC0-1.0 | runbook | inspired-by |
Scoutflo/Scoutflo-SRE-Playbooks @ acc55da0224f | MIT | runbook | inspired-by |
WynterJones/CoppieGPT @ 96fc78149f03 | NONE | muster-author | inspired-by |
boraoztunc/skills @ 1209fb6b07b8 | Apache-2.0 | muster-author | inspired-by |
Works
Articles, books, papers, standards, talks and people whose ideas, frameworks or structure an item is based on. No text is carried from them.
| Work | Kind | Basis for |
|---|---|---|
| How to Write a Novel Using the Snowflake Method, Randy Ingermanson | book | book-run |
| Unlock AI: Open Engine and Open Skills, Nate B. Jones | person | review-gate, muster-humanizer, muster-image, muster-research, prd, case-study, newsletter, blog-post, social-post |
| Building effective agents, Erik Schluntz and Barry Zhang (Anthropic) | article | muster-runner |
| Create custom subagents (Claude Code docs), Anthropic | article | muster-runner |
| LLM-Blender: Ensembling Large Language Models with Pairwise Ranking and Generative Fusion, Dongfu Jiang, Xiang Ren and Bill Yuchen Lin (ACL 2023) | paper | tournament |
| Prompting best practices (Claude Docs), Anthropic | article | muster-prompt-smith |
| Reduce hallucinations (Claude Docs, Strengthen guardrails), Anthropic | article | muster-prompt-smith |
| Define success criteria and build evaluations (Claude Docs), Anthropic | article | muster-prompt-smith, muster-pm |
| Demystifying evals for AI agents, Anthropic | article | muster-pm |
| Evaluation best practices (OpenAI API docs), OpenAI | article | muster-pm |
| Your AI Product Needs Evals, Hamel Husain | article | muster-pm |
| AI Evals: Everything You Need to Know, Hamel Husain and Shreya Shankar | article | muster-pm |
| The new rules of context engineering for Claude 5 generation models, Anthropic | article | muster-prompt-smith |
| LLM-Cite | paper | muster-research |
| Wikipedia:Signs of AI writing, WikiProject AI Cleanup (Wikipedia editors) | article | muster-humanizer |
| AIDA (Attention, Interest, Desire, Action), E. St. Elmo Lewis | person | muster-author, social-post, lead-magnet |
| The QUEST copywriting formula, Michel Fortin | article | muster-author, social-post |
| How to Write Copy That Sells (the PASTOR framework), Ray Edwards | book | muster-author |
| Breakthrough Advertising (awareness and sophistication), Eugene M. Schwartz | book | muster-author |
| Search Quality Rater Guidelines (E-E-A-T), Google | standard | muster-author, blog-post |
| Creating helpful, reliable, people-first content, Google Search Central | article | blog-post |
| GEO: Generative Engine Optimization, Pranjal Aggarwal et al. (KDD 2024) | paper | blog-post |
| AICMO marketing prompt collection; MaxsPrompts | article | competitive-battlecard, lead-magnet, case-study, newsletter |
| Sales Battlecards 101, Klue | article | competitive-battlecard |
| Modern battlecard blueprint, Crayon | article | competitive-battlecard |
| Challenge-Solution-Results case study structure, libril.com and geisheker.com | article | case-study |
| Email deliverability guidance, Litmus and Validity | article | newsletter |
| The Hero with a Thousand Faces, Joseph Campbell | book | case-study |
| Building a StoryBrand (SB7), Donald Miller | book | book, case-study |
| The Pyramid Principle (SCQA, MECE), Barbara Minto | book | executive-summary, book |
| Save the Cat!, Blake Snyder | book | book, book-run |
| The Story Grid, Shawn Coyne | book | book, book-run |
| Techniques of the Selling Writer (scene and sequel), Dwight V. Swain | book | book, book-run |
| Chapter Blog nonfiction outline; ProWritingAid non-fiction structure | article | book, book-run |
| Competing Against Luck (Jobs to Be Done), Clayton M. Christensen, Taddy Hall, Karen Dillon and David S. Duncan | book | muster-author, prd, epic |
| Lean UX (hypothesis statements, proto-personas), Jeff Gothelf and Josh Seiden | book | prd, epic |
| The Humanizing Work Guide to Splitting User Stories, Richard Lawrence and Peter Green | article | epic |
| User Story Mapping, Jeff Patton | book | epic |
| Working Backwards (PR/FAQ), Colin Bryar and Bill Carr | book | launch-plan |
| Semantic Versioning 2.0.0, Tom Preston-Werner | standard | release-notes |
| Measure What Matters (OKRs), John Doerr | book | okrs |
| High Output Management, Andrew S. Grove | book | okrs |
| RICE: simple prioritization for product managers, Sean McBride (Intercom) | article | roadmap, muster-roadmap |
| The Now-Next-Later roadmap, Janna Bastow (ProdPad) | article | roadmap, muster-roadmap |
| Documenting Architecture Decisions, Michael Nygard | article | design |
| ISO/IEC/IEEE 29119-3, Software testing: Test documentation, ISO/IEC/IEEE | standard | ai-test-plan |
| IEEE 829, Standard for Software and System Test Documentation (withdrawn), IEEE | standard | ai-test-plan |
| Site Reliability Engineering (Managing Incidents), Betsy Beyer, Chris Jones, Jennifer Petoff and Niall Richard Murphy (Google) | book | runbook |
| PagerDuty Incident Response documentation, PagerDuty | article | runbook |
| Runbook practice articles (preconditions, observability, escalation, cleanup, idempotency), emmer.dev, The Good Shell and Rootly | article | runbook |
| 10 Usability Heuristics for User Interface Design, Jakob Nielsen (Nielsen Norman Group) | article | audit-pattern-design-ux |
| Timeless documentation (Google developer documentation style guide), Google | article | audit-pattern-documentation |
| Diátaxis, Daniele Procida | article | audit-pattern-documentation |
| Cognitive Complexity, Because Testability != Understandability, G. Ann Campbell (SonarSource) | article | audit-pattern-readability |
| OWASP Top 10:2021 A08, Software and Data Integrity Failures, OWASP Foundation | standard | audit-pattern-security |
Origins not yet identified
Muster's history says these items drew on sources it never named. Each note records what was searched.
| What | Note | Basis for |
|---|---|---|
| PAS (Problem-Agitate-Solution) and BAB (Before-After-Bridge) copywriting formulas | 2026-11-09: named in muster-author, social-post, lead-magnet and book. Neither muster's history nor a primary source found on this date names who originated either formula. | muster-author, social-post, lead-magnet, book |
| The "high-signal sources" behind the first content rubrics | 2026-11-09: commit b99599a (2026-06-07) says the blog, social, lead-magnet, newsletter, case-study and runbook rubrics were "grounded in AIDA/PAS/BAB/QUEST + E-E-A-T from high-signal sources" without naming them; no doc was committed with it. The 2026-06-08 research note (b122ed9) names later sources for most of these pipelines, recorded here as their own entries; the original set is not recoverable. | blog-post, social-post, lead-magnet, newsletter, case-study, runbook |
| BLUF (bottom line up front) | 2026-11-09: named in executive-summary. A military-correspondence convention; no single originating work was identified on this date. | executive-summary |
License review
Some sources carry no license, a NonCommercial or NoDerivatives license, or a proprietary one, so Muster has no right to reuse their text. Each was compared with the Muster items based on it, and the manifest records the result as a licenseReview: ideas-only when no text was carried, or clean-room-rewrite when Muster text that followed the source too closely was rewritten from the underlying idea (on 2026-11-09: the humanizer's vocabulary lists, the PRD section structure, the epic hypothesis template, and five quotations in the audit hunt lists, now paraphrased with their links kept). The test suite fails on any such source without a review, and on any waiver still pending a decision.
Upstream routing on Codex
On Codex, Muster prefers an enabled authoritative upstream implementation over its own copy whenever one exists: the official Superpowers plugin, wshobson's per-plugin Codex skills, and GSD's installer-generated Codex skills.
The bundled sp-*, wsh-*, and muster-gsd-* skills are deterministic fallbacks: they keep every role resolvable when an upstream is absent, and they never implicitly install those providers on your behalf. The pinned compatibility survey — which upstreams were checked, at which refs, including Atomic Codex, Book Genesis, the humanizer sources, and Promptfoo — is recorded in codex/upstreams.json.
Clean-room specialists
Alongside the vendored material, Muster ships its own specialists in plugin/agents/, authored fresh from atomic-claude's role concept:
- muster-surgeon: precise 1-2 file edits
- muster-investigator: read-only locator
- muster-strategist: heavyweight reasoning
- muster-improver: read-only post-run retrospective that proposes self-improvement edits
The muster-runner agent (built from the lifecycle mechanisms credited above) left the Claude plugin in revival stage 2b. It survives only as the frozen Codex lane's production-wave policy, in codex/frozen/agents/muster-runner.md.
Lifecycle seats adapted from GSD
The stage-2 lifecycle agents are adapted from gsd-core v1.20.1 agents (MIT). gsd's measured protocol text (honest abstain, the evidence gate, revision-loop limits, failing direction, the must-NOT probe, deviation rules, the self-check and the destructive-git ban) is carried verbatim in marked regions; muster vendor update gsd keeps them in sync by 3-way merge:
- muster-planner: writes the run's PLAN.md (from gsd-planner)
- muster-plan-checker: checks PLAN.md before execution (from gsd-plan-checker)
- muster-executor: implements one task in its own worktree (from gsd-executor, with the former builder's one-slice scope contract)
- muster-verifier: goal-backward verification with honest abstain (from gsd-verifier)
- muster-reviewer: adversarial review of one task's commits (from gsd-code-reviewer, with muster's verdict discipline)
- muster-test-writer: writes a task's failing tests before anyone implements it (from gsd-nyquist-auditor)
- muster-integration-checker: checks that a wave's parallel tasks wire together (from gsd-integration-checker)
The muster:debug workflow's hypothesis investigator, wsh-team-debugger, is adapted from wshobson's agent-teams team-debugger (MIT).
Additional inspiration (clean-room, mechanism-level)
A cluster of systems introduced in 0.5.0 -- the sprint coordination protocol (claim/receipt/ledger discipline), the citation-guard gate, QA runbook memory, the voice/audience/brand profile system, newsletter signal-diff, the assumption checker, and the document-ingestion contract -- were designed with mechanism-level inspiration from Nate B. Jones's Unlock AI open-skills/open-engine patterns (unlock-ai.natebjones.com). No code, prose, or configuration was copied; Muster's implementations are original, adapted to its own deterministic-CLI/model-facing-skill architecture. The Unlock AI library is members-only, so this could not be checked against its text; the items it informed are listed under Works above.
License
Muster is licensed under Apache-2.0. See LICENSE and NOTICE.
Contributing
Contributions are welcome. See CONTRIBUTING.md to get started.