By Y4tacker
Claude Code plugin that supervises durable Fact-Intent blackboard runs through an internal waymark CLI.
Attempts a direct first-pass solve for a Waymark run using only the waymark CLI blackboard.
Claims exactly one Waymark intent and concludes one incremental fact.
Reviews a Waymark Fact-Intent graph, creates next intents, or completes through the CLI.
Re-verifies Waymark completion evidence before the main session prints WAYMARK_RUN_COMPLETE.
Executes bash commands
Hook triggers when Bash tool is used
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.
Executables (bin/) — files in this plugin's bin/directory are added to the Bash tool's PATH while the plugin is enabled.
Waymark is a Claude Code plugin for running long-horizon agent work through a persistent, SQLite-backed blackboard instead of a fragile chat transcript. A user-approved mission contract seeds a Fact-Intent graph; Claude subagents then coordinate indirectly by writing durable facts, intents, claims, releases, completion evidence, and verification records through the waymark CLI.
The CLI and blackboard.sqlite are the authority. Generated files such as STATE.md, graph.yaml, timeline.txt, events.jsonl, hook-events.jsonl, and reports/final.md are views and logs for humans and agents. A run is complete only when waymark final-status --json returns ready=true.
Waymark is domain-general. It can drive software work, research, writing, data analysis, planning, or any task where the final state can be described as evidence-backed criteria rather than a vague "done" claim.
Long-running agent work has a state problem:
Waymark separates mission, coordination, execution, and completion proof:
waymark-goal drafts a mission contract: origin, goal, falsifiable acceptance criteria, constraints, assumptions, and handoff triggers.waymark init; init is the semantic commitment point.blackboard.sqlite./goal + PROTOCOL.md supervises the run and dispatches subagents, but it does not own semantic state.bin/waymark.audit, verify, verification-record, and final-status decide whether completion is real.flowchart TD
U["User request"] --> WG["waymark-goal skill"]
WG --> MC["Mission contract"]
MC --> INIT["waymark init"]
INIT --> DB[("blackboard.sqlite")]
DB --> SUP["/goal + PROTOCOL.md"]
SUP --> BOOT["bootstrap-worker"]
SUP --> REASON["reason-worker"]
SUP --> EXPLORE["explore-worker"]
BOOT --> CLI["bin/waymark"]
REASON --> CLI
EXPLORE --> CLI
CLI --> DB
DB --> VIEWS["Generated views"]
VIEWS --> STATE["STATE.md"]
VIEWS --> GRAPH["graph.yaml"]
VIEWS --> TIME["timeline.txt"]
VIEWS --> EVENTS["events.jsonl"]
VIEWS --> REPORT["reports/final.md"]
SUP --> AUDIT["audit"]
AUDIT --> VERIFIER["verifier-worker"]
VERIFIER --> RECORD["verification-record"]
RECORD --> FS["final-status"]
FS -->|ready=true| DONE["WAYMARK_RUN_COMPLETE"]
FS -->|handoff or failed proof| HANDOFF["WAYMARK_HANDOFF"]
HOOKS["PostToolUse / SubagentStop / Stop hooks"] -. "observe; strict Stop can gate" .-> FS
WF["Dynamic Workflow reference"] -. "optional bounded waves" .-> SUP
Waymark is built around explicit principle/mechanism pairs.
npx claudepluginhub y4tacker/waymarkStructure agent work as Mission, Task Contract, Evidence, User Judgment, and reflection.
Advanced Hive Mind collective intelligence system for queen-led multi-agent coordination with consensus mechanisms and persistent memory
AI agent framework for building autonomous workflows. Enables no-code creation of intelligent task automation for nonprofit operations.
Markdown-only Agent Team workflow for Claude Code without requiring a Mexus server.
Proactive enhancement layer for AI agents — learns user patterns from any memory store, upgrades prompts into full intent, takes engineer-grade initiative that advances the goal instead of generic filler, predicts the next request, verifies everything before delivery, scans for AI slop, orchestrates subagents when available, onboards itself to any host on first run, and runs a self-improvement loop that sharpens its own heuristics. Universal adapter runs on Claude Code, Codex, Hermes, OpenClaw, opencode, Cursor, Gemini CLI, and custom loops.
Grounded AI behavior: named responses and living task documents for seamless context handoff