By bburda
Rationale-as-code layer for sphinx-needs: AI-agent decisions, risks, and open questions as first-class needs linked to requirements with impact analysis across the decision graph
Atomic — mark a memory deprecated, optionally with supersedes link
Atomic — list memories whose review_after has passed
Atomic — fetch a single memory in full by id
Atomic — bidirectional graph walk from a starting need
Atomic — create a Papyrus workspace at a given path
Uses power tools
Uses Bash, Write, or Edit tools
Own this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimOwn this plugin?
Verify ownership to unlock analytics, metadata editing, and a verified badge. GitHub access is read-only (username + org membership).
Sign in to claimBased on adoption, maintenance, documentation, and repository signals. Not a security audit or endorsement.
Rationale-as-code layer for sphinx-needs and Pharaoh.
Papyrus stores AI-agent decisions, risks, facts, and open questions as
first-class sphinx-needs, linked to your requirements via the same link
types Pharaoh already uses (satisfies, derives, extends,
supersedes). The differentiator: papyrus impact <req-id> walks the
memory↔requirement graph and returns the full chain of decisions and
risks affected by a change request.
git clone https://github.com/useblocks/papyrus.git
cd papyrus
uv venv
uv pip install -e .
source .venv/bin/activate
Verify:
papyrus --version
uv pip install -e ".[semantic]"
Adds the all-MiniLM-L6-v2 model (English-only, ~90 MB on first use).
Enables papyrus recall --semantic -q "..." for similarity search over
titles, bodies, and tags.
papyrus init /tmp/memory-ws
papyrus --workspace /tmp/memory-ws add dec "Use bcrypt for password hashing" --confidence high
papyrus --workspace /tmp/memory-ws recall --format brief
The full flow — linking to an external pharaoh requirement, running
impact analysis, tracing a decision's history — is walked through in
docs/tutorial.md.
Papyrus ships an MCP server with 12 tools (recall, add, link, impact, trace, …). Wire it into your client:
Claude Code:
claude mcp add papyrus -- papyrus mcp-serve
GitHub Copilot (VS Code) — add to .vscode/mcp.json:
{
"servers": {
"papyrus": {
"command": "papyrus",
"args": ["mcp-serve"],
"cwd": "${workspaceFolder}"
}
}
}
Both are detailed in docs/mcp-setup.md with
verification steps and troubleshooting.
papyrus impact <req-id>
traverses the memory↔requirement graph in both directions — one
command replaces a lot of grepping.docs/tutorial.md — hands-on 10-minute walkthroughdocs/mcp-setup.md — Claude Code + Copilot integrationdocs/architecture.md — module layout and data modeluv pip install -e ".[dev]"
pytest -v
ruff check src tests
pyright
All three must pass on Python 3.10, 3.11, 3.12 (CI enforces). See
CONTRIBUTING.md.
MIT — see LICENSE.
npx claudepluginhub bburda/papyrus --plugin papyrusAI assistant framework for sphinx-needs projects: change analysis, traceability, MECE, authoring, verification, and release management
Core knowledge management, documentation generation, and strategic analysis for Claude Code
You work with me (Claude) - I guide your workflow and suggest next actions.
Comprehensive AI-assisted requirements elicitation. Supports stakeholder interviews (LLMREI pattern), document extraction, stakeholder simulation, domain research, gap analysis, user story mapping, customer journey mapping, JTBD analysis, prioritization (MoSCoW/Kano/WSJF), surveys, workshops, brainstorming, and business rules analysis. Exports to canonical, EARS, and Gherkin formats.
ARIA — Applied Reasoning and Insight Architecture. Persistent human-governed knowledge for Claude Code via a five-phase lifecycle: capture → govern → promote → apply → refresh. Stages session insights/decisions/feedback into review backlogs, promotes approved items to a tag-indexed markdown base, and applies them via /context, /rules, /codemap, /stitch, /distill, /prospect (forward-looking pre-mortems on plans before execution), /retrospect (commit/range/PR/release/deployment retrospectives with per-fix validation), and Rule 22 change-decision hooks on every Edit/Write. Both /prospect and /retrospect run an Evidence-Sourcing Pass that autonomously sources accessible evidence (codebase reads, public docs, MCP queries) and surfaces user-input asks for anything that requires judgment — converting unsupported assumptions to validated/falsified before the report finalizes. Reports persist to ~/knowledge/logs/{prospect,retrospect}/ with structured frontmatter and become discoverable via /context. Ships 34 working rules + a 7-step framework, audit cadences, ideas-lifecycle routing, drift detection, an optional shared-knowledge tier for team promotions, and structural-signal surfacing on critical paths. IMPORTANT: All skills except /setup require ~/.claude/aria-knowledge.local.md to exist. If missing when any skill is invoked, stop and tell the user to run /setup.
Upstash Context7 MCP server for up-to-date documentation lookup. Pull version-specific documentation and code examples directly from source repositories into your LLM context.