Magnus HedemarkandGitHub
e10508b034
* feat(bmad): add BMad control-plane protocol skill New standalone methodology skill that lets any agent run the BMad method (Breakthrough Method of Agile AI-Driven Development) as a harness-agnostic control-plane protocol: five-field intent contracts, direct/bounded/initiative classification, review-as-triage, failure routing by layer, and autonomy gating with machine-readable spec status. - SKILL.md protocol core with progressive disclosure + When not to use - README.md human-facing install guide - 9 references: protocol, classification, spec, lifecycle, project-context, review-and-failure-routing, autonomy, party-mode, adoption - 4 templates: SPEC, INTENT, STORY, REVIEW - scripts/check-spec.py + 16 tests (stdlib, deterministic spec validation) - evals/evals.json: 9 output-quality cases - Routing seams from bmad to adjacent skills and back from spec-driven-development, product-shaping, implementation-planning, neckbeard - Catalog updates: root README, skill-triggers, marketplace/plugin/llms.txt Closes #399 * fix(bmad): address droid-review findings - check-spec.py: skip headings inside fenced/indented code blocks so a spec cannot PASS on section text that only appears in a code sample - check-spec.py: catch UnicodeDecodeError on non-UTF-8 files and report FAIL instead of crashing - STORY.md template: add created key for resumability/traceability parity - SPEC.md template: split in-progress and in-review status bullets - add 2 regression tests (heading-in-fence, non-UTF-8) * fix(bmad): address droid-review round 2 - check-spec.py: read specs with utf-8-sig so a UTF-8 BOM cannot silently disable the frontmatter status check - check-spec.py: handle standard YAML inline comments after status values (status: draft # pending review) without a false FAIL - references/protocol.md: make lifecycle phrasing consistent with lifecycle.md — four phases plus a learning closeout - add 2 regression tests (BOM, inline comment) * fix(bmad): tolerate trailing whitespace on frontmatter delimiters A spec whose --- delimiter lines carry trailing spaces or tabs would silently disable the status check and let an invalid status PASS. Relax the delimiter pattern and add a regression test. * fix(bmad): ignore inline comments in quoted status values * fix(bmad): tolerate leading blank lines before frontmatter * fix(bmad): fail closed on unparseable frontmatter, matching fence markers Address droid-review round 5 and 6 findings as a single closed class: - Fail closed when a file opens with a --- delimiter that cannot be parsed, so no whitespace/frontmatter permutation can silently disable the status check (previously: unparseable frontmatter was treated as 'no status' warning, letting an invalid status PASS). - Track fence opener markers in collect_headings so a mismatched fence no longer closes a code block early (false-PASS on missing sections) and an unclosed fence no longer swallows real headings. - Accept empty well-formed frontmatter (---\n---) and closing delimiters without a trailing newline. - STORY.md template: parent-spec points at the sibling SPEC.md. - README: status vocabulary is not a strict linear chain; blocked is a resumable routing signal. Whitespace/frontmatter mutation sweep: 9 formatting variants x valid/invalid status all verdict correctly; malformed delimiters fail closed. 29 tests.
product-shaping
Set appetites instead of estimates, shape raw ideas into bounded pitches, and bet on them with capped downside — Basecamp's Shape Up method, adapted for teams whose builders include AI agents.
Why Install This Skill
Most failed work doesn't fail during the build. It fails before it starts: an idea enters the pipeline with no one having decided what it's worth, gets built to whatever size the first design implies, and drags past every deadline because nothing defined where it stops. Teams respond with estimation rituals that produce numbers nobody believes, or backlogs that grow until nobody reads them.
This skill installs the alternative: a pre-commitment discipline that converts raw ideas into bounded pitches (problem, appetite, solution, rabbit holes, no-gos), bets a fixed time budget on them, and governs the build with discovered scopes and deliberate scope cutting. It is adapted from Ryan Singer's Shape Up (Basecamp, 2019) and extends the method for a reality the book never faced: build teams that include AI agents — where budgets are measured in sessions and cost ceilings, review time is a first-class scope cost, and non-converging fix loops need explicit kill criteria.
After installing, an agent can take "customers want a calendar" and return a bounded pitch for the tenth of a calendar worth building; take a "redesign X" grab-bag and either re-anchor it on a specific problem or route it to proper discovery; run a build that integrates one real slice early instead of assembling disconnected parts at hour eleven; and stop a review/fix spiral at round three instead of round nine.
What You Get
| Directory | What it provides |
|---|---|
SKILL.md |
Core loop and entry-point routing: shape → bet → build → move on |
references/principles.md |
Appetite vs estimates, fixed-time-variable-scope, the rough/solved/bounded test, evidence boundaries, method lineage |
references/shaping.md |
The four shaping steps in detail, including breadboarding and shaping non-UI work (APIs, infrastructure, agent workflows) |
references/betting.md |
Bets vs backlogs, circuit-breaker mechanics, cycle length as a tunable parameter, defect handling between bets |
references/building.md |
Hand-over-responsibility, one-piece-done integration, scope mapping, uphill/downhill progress tracking, deciding when to stop |
references/hybrid-adaptation.md |
The agent-team extension: budget currencies, batched steering, verification cost inside scope, kill criteria for non-converging loops, earned autonomy |
references/anti-patterns.md |
Documented field failures (2019–2026 independent team records) and agent-workflow failure modes, with mitigations |
templates/PITCH.md |
Fillable five-ingredient pitch document |
templates/SCOPE_MAP.md |
Fillable scope table with hill states and circuit-breaker check |
evals/evals.json |
Output-quality evaluation cases |
Triggers
Load this skill when:
- A raw idea, feature request, or "redesign/improve X" request needs to become a bounded project before work starts
- Someone asks "how much should we invest in this?" or "what's the smallest version worth doing?"
- Planning a delegated AI-agent build: budgets, verification cost, steering cadence, or kill criteria for loops that won't converge
- A project keeps not finishing and the honest question is whether it was ever shaped
- An epic resists decomposition because nobody can define done
- Post-ship feedback is threatening to become an instant commitment
Requirements
No software dependencies or credentials. Reference-only skill: it shapes thinking and documents, not systems.
Quick Start
No setup needed. Trigger examples:
- Say "help me shape this idea before we commit time to it" or "set an appetite for
this work" — start at
references/shaping.md - Say "write a pitch for this" — fill in
templates/PITCH.md - Say "we're delegating this build to an agent, scope it" — load
references/hybrid-adaptation.mdfor budgets and kill criteria - Say "this project won't stop running long / this loop won't converge" — load
references/anti-patterns.mdfirst, then hybrid-adaptation kill criteria
Pairs naturally with product-discovery (upstream problem validation),
product-methodology (downstream prioritization of a won bet), and delegation/kanban
skills when builders are agents.
Attribution
Method adapted from Shape Up by Ryan Singer (Basecamp, 2019), available free at basecamp.com/shapeup. This skill is an original distillation with independent field evidence and an original hybrid human+AI adaptation layer, not a reproduction of the book.