42 tagged dag, measured the same way as everything else here.
Browse within: agent-orchestration 22pi-coding-agent 18subagents 18taskflow 18
Agent
Golf Lean 4 proofs after they compile; improve proofs for directness, clarity, performance, and brevity without changing semantics. Use after successful compilation to achieve 30-40% size reduction.
Agent
Compiler-guided iterative proof repair with two-stage repair escalation (fast → strong). Use for error-driven proof fixing with small sampling budgets (K=1).
Agent
Strategic resolution of stubborn sorries; may refactor across files within the header fence. Use when fast pass fails or for complex proofs.
Agent
An agent that chooses a main workflow template and combines cross-domain safeguards into one YAML DAG, a file describing dependent tasks.
Agent
How engineering skills consume this repository's domain documentation while exploring the codebase.
Agent
Issues and PRDs for this repo live as GitHub issues. Use the gh CLI for all operations.
Agent Claude Code
Analyze blind comparison results to understand WHY the winner won and generate improvement suggestions.
Agent Claude Code
Compare two outputs WITHOUT knowing which skill produced them.
Agent Claude Code
Evaluate expectations against an execution transcript and outputs.
Agent
Author and edit documentation FILES on disk (README, guides, changelogs, docs).
Agent
Reviews code, plans, architecture risk, and test gaps without editing files.
Agent
Review changes for security vulnerabilities and trust-boundary issues.
Agent
Read this before dispatching agents. Every rule here was paid for. project/agents/jobs.md covers the tracker the work is filed in; project/agents/harness.md covers the tool the agents run inside.
Agent
Facts about the tool an agent runs inside, not about this project — they are identical in every repo, which is why they are not in project/gotchas.md. One section per harness; today there is only Claude Code. Read this the first time a call fails with a permission error or a refusal you did not expect: none of these…
Agent
Jobs is the tracker: where work is filed, how big a unit of work is, and what a fan-out mints before it launches. project/agents/delegation.md covers running agents; this file covers the tracker they claim from. job --help is the authority on flags — this file is the house convention around them.
Agent
Composition uses ordinary application boundaries. Junjo does not add a generic Agent Node, Workflow Tool, shared Store mapper, or universal executable base.
Agent
Agent is a reusable, typed Junjo executable for the case where a model chooses the next capability at runtime. It is a sibling of Workflow. An Agent has no fabricated Graph, and a Tool is an operation rather than a Node.
Agent
junjo.agent.testing is the public deterministic test boundary. It has no provider dependency.
Samuel0101010/wisp-orchestrator
Agent
Invoke to implement exactly one task from tasks.md. Reads architecture.md as a binding constraint, writes code, and stops after one logical commit.
Samuel0101010/wisp-orchestrator
Agent
Invoke to convert a Goal plus a Team specification into a valid Plan DAG written to plan.json. Used once per harness run before agents are dispatched.
Samuel0101010/wisp-orchestrator
Agent
Invoke after a developer commit to run build, test, and lint and emit a BMAD-style PASS/CONCERNS/FAIL gate verdict as JSON. Read-only; never modifies code.
Agent
Offloads conversation context into ForkMind's encrypted, immutable DAG capsules so the live model window can be compacted without losing anything. Spawn when context is getting full, when the user says "save this and clear it", "archive the earlier discussion", "offload context", or when a large block (file dumps…
Agent
Drives ForkMind to debug, compare, or regression-test LLM / agent calls in an isolated context. Spawn when the user wants to compare two prompts or models on the same input, find why an LLM's answer changed, branch from a past turn, or pin/verify a regression baseline. Returns a compact verdict (winner, diffs, drift)…