oss-a-001record 1
{
"rank": 1,
"id": "oss-a-001",
"identity": "github/spec-kit",
"url": "https://github.com/github/spec-kit",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "NEEDS-CLARIFICATION marker written into the artifact at the point of ambiguity; spec gateable on zero open markers",
"mechanism": "Markdown templates under templates/ (spec-template.md, plan-template.md, tasks-template.md, constitution-template.md, checklist-template.md); project artifacts under specs/ and .specify/ (templates, overrides, presets, extensions, memory). Ambiguity is encoded IN the artifact as literal '[NEEDS CLARIFICATION: <question>]' markers.",
"limitations": "Template structure verified from source; I did NOT execute the CLI, so real-world output quality/adherence of any given agent to the template is unobserved. Star count (131k) reflects hype cycle, not deployed quality.",
"license_claimed": "MIT (GitHub API spdx MIT; consistent with README statement)",
"disposition": "top10_promoted"
}
oss-a-027record 2
{
"rank": 2,
"id": "oss-a-027",
"identity": "cyanheads/survey-mcp-server",
"url": "https://github.com/cyanheads/survey-mcp-server",
"segment": "dialog",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Server owns question state/eligibility/validation, LLM owns phrasing only; eligibilityReason makes wrong skips debuggable",
"mechanism": "VERIFIED FROM SOURCE (survey-definitions/product/feedback/comprehensive-2025.json, 21 questions): top-level {id, version, metadata, questions, settings}. Each question = {id, type (email|multiple-choice|...), text, helpText, required, options[{value,label}]}. Conditional logic is an explicit boolean tree: \"conditional\": {\"operator\":\"OR\", \"conditions\":[{\"dependsOn\":\"user_role\",\"showIf\":[\"team-admin\"]}]}.",
"limitations": "Only 4 stars and 10 months idle — this is a design reference, NOT a dependency to adopt. I read one definition file; I did not run the server or verify the eligibility engine actually honours the conditional tree at runtime.",
"license_claimed": "Apache-2.0 (GitHub API spdx; matches README badge)",
"disposition": "top10_promoted"
}
oss-a-002record 3
{
"rank": 3,
"id": "oss-a-002",
"identity": "Fission-AI/OpenSpec",
"url": "https://github.com/Fission-AI/OpenSpec",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Change-scoped ADDED/MODIFIED/REMOVED deltas plus a stakes-free explore phase before committed scope",
"mechanism": "openspec/changes/<change-name>/ containing proposal.md, specs/, design.md, tasks.md; archived to openspec/changes/archive/<YYYY-MM-DD>-<name>/; long-lived specs in openspec/specs. Delta syntax is Markdown: '## ADDED Requirements' > '### Requirement:' (SHALL language) > '#### Scenario:' with WHEN/THEN bullets.",
"limitations": "Directory/format read from README; I did not run openspec init to confirm generated layout matches docs. The 'superior token efficiency' claim vs BMAD is vendor/blog framing, not something I measured.",
"license_claimed": "MIT (GitHub API spdx MIT)",
"disposition": "top10_promoted"
}
oss-a-005record 4
{
"rank": 4,
"id": "oss-a-005",
"identity": "gokulrajaram/ProductSpec",
"url": "https://github.com/gokulrajaram/ProductSpec",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Only formally schematised spec format found; artifact_type enum encodes a maturity ladder; decision-trace as first-class object",
"mechanism": "'.product-spec.md' files: YAML frontmatter + '## Section Label' Markdown sections, minItems 6. Frontmatter requires spec_format_version/title/artifact_type/author/created_at/updated_at; artifact_type is an ENUM of hypothesis|prd|openspec_proposal. Formal JSON Schemas in schema/: product-spec.schema.json, agent-run.schema.json, decision-trace.schema.json, review-annotation.schema.json.",
"limitations": "I read SPEC.md and the JSON schema directly. Low adoption (275 stars) — a standard with no ecosystem is a proposal, not an interop layer. Conformance suite exists but I did not run it.",
"license_claimed": "MIT (GitHub API spdx MIT)",
"disposition": "top10_promoted"
}
oss-a-028record 5
{
"rank": 5,
"id": "oss-a-028",
"identity": "oasis-surveys/oasis-platform",
"url": "https://github.com/oasis-surveys/oasis-platform",
"segment": "dialog",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Shadow mode: adaptive policy non-acting by default, logging what it would have asked for human review (AGPL — pattern only)",
"mechanism": "backend/app/schemas/ contains agent.py, analytics.py, participant.py, session.py, study.py — study/agent/session is the domain model. Four built-in templates (semi-structured qualitative, cognitive interview pretest, open-ended survey follow-up, telephone survey).",
"limitations": "I confirmed the schema MODULE names but did not read the guide field definitions, so the exact probe/transition representation is unverified. Notable HITL design: the adaptive engagement policy ships in NON-ACTING SHADOW MODE, logging what it would have done to an audit trail for researcher review before going live.",
"license_claimed": "BADGE TRAP — API reports NOASSERTION; LICENSE body reads 'SPDX-License-Identifier: AGPL-3.0-only'. Deploying a MODIFIED version as a network service forces disclosure of changes. Hard blocker for proprietary client SaaS.",
"disposition": "top10_promoted"
}
oss-a-032record 6
{
"rank": 6,
"id": "oss-a-032",
"identity": "surveyjs/survey-library",
"url": "https://github.com/surveyjs/survey-library",
"segment": "form-engine",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Headless survey-core plus swappable renderers; carry-forward and piping prevent the didn't-you-just-ask-me-that failure",
"mechanism": "SurveyJS JSON: pages -> elements/questions with 20+ built-in types; conditional expressions (visibleIf/enableIf-family), calculated values, and triggers all declared IN the JSON so logic travels with the definition.",
"limitations": "I verified the package split and read the Creator EULA, but did NOT read raw schema source to quote exact property names, so specific keys are from docs/description rather than code. This MIT-renderer / paid-builder split is the Composio pattern again: the badge is true but covers less than it appears.",
"license_claimed": "MIT for the LIBRARY (verified: packages/ = survey-core + 4 UI adapters, spdx MIT). BUT the drag-drop builder is a SEPARATE repo, surveyjs/survey-creator, whose LICENSE body is a 'Commercial developer license ... Devsoft Baltic OU ... END-USER LICENSE AGREEMENT'. Renderer free; authoring UI paid.",
"disposition": "top10_promoted"
}
oss-a-045record 7
{
"rank": 7,
"id": "oss-a-045",
"identity": "cucumber/cucumber-js",
"url": "https://github.com/cucumber/cucumber-js",
"segment": "executable-spec",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Two-audience artifact: one document a client approves and a machine executes as a pass/fail gate",
"mechanism": "Gherkin .feature files (Feature:/Scenario:/Given/When/Then) in features/, bound to step definitions in features/support/steps.js via pattern matching with typed placeholders like {string}; shared state via the 'World' (this). Run with npx cucumber-js.",
"limitations": "Verified format and binding from README examples. Cucumber CANNOT drive a UI alone — it needs Playwright/Selenium/Cypress underneath. The well-known failure mode (step-definition maintenance overhead, Gherkin written by devs not stakeholders) is not visible from the repo.",
"license_claimed": "MIT (GitHub API spdx MIT)",
"disposition": "top10_promoted"
}
oss-a-003record 8
{
"rank": 8,
"id": "oss-a-003",
"identity": "bmad-code-org/BMAD-METHOD",
"url": "https://github.com/bmad-code-org/BMAD-METHOD",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "D",
"verified": "fetched",
"denominator": 60,
"selection_reason": "Role-differentiated interrogation usable as a question-taxonomy coverage axis without importing the machinery",
"mechanism": "src/ contains only bmm-skills/, core-skills/, scripts/ — v6 refactored personas into 'skills' rather than flat agent .md files. Produces briefs, PRDs, architecture docs, and story files.",
"limitations": "I confirmed the src/ top level and LICENSE body but did NOT descend into bmm-skills to read actual persona/elicitation prompt text, so the concrete question-asking mechanism is unverified. Widely reported as high-token and slow (one blog claims 5.5h vs 12min vs OpenSpec) — that benchmark is third-party and unreplicated.",
"license_claimed": "API reports NOASSERTION; I read LICENSE body: it IS 'MIT License, Copyright (c) 2025 BMad Code, LLC' but with added contributor-attribution prose, plus a separate TRADEMARK.md reserving 'BMad'/'BMAD-METHOD' marks. So: MIT code, restricted naming.",
"disposition": "top10_promoted"
}
oss-a-018record 9
{
"rank": 9,
"id": "oss-a-018",
"identity": "Priivacy-ai/spec-kitty",
"url": "https://github.com/Priivacy-ai/spec-kitty",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "I",
"verified": "search-only",
"denominator": 60,
"selection_reason": "Retained to document spec-driven duplication: near-identical fork lineage with murky provenance",
"mechanism": "Spec tasks surfaced as Kanban cards backed by git worktrees.",
"limitations": "Not fetched. Interesting only for the spec-as-visible-board UX idea.",
"license_claimed": "MIT (GitHub API spdx MIT)",
"disposition": "top10_promoted"
}
oss-a-025record 10
{
"rank": 10,
"id": "oss-a-025",
"identity": "leocamello/spec-kit-v-model",
"url": "https://github.com/leocamello/spec-kit-v-model",
"segment": "spec-driven",
"observed": "2026-08-27",
"evidence_class": "I",
"verified": "search-only",
"denominator": 60,
"selection_reason": "V-model variant retained to show the segment's crowding rather than padding it as distinct mechanism",
"mechanism": "Paired dev-spec/test-spec artifacts layered on spec-kit templates.",
"limitations": "Not fetched; tiny. The PAIRING RULE is the transferable idea.",
"license_claimed": "MIT (search metadata)",
"disposition": "top10_promoted"
}