Bootstrap a DPD (Decompose-Propagate Decision) session under the ambient overlay paradigm — DPD is a passive graph overlay that observes ongoing conversation and proposes graph updates collaboratively. Invoke when the user runs /dpd, or proactively suggest when a conversation has accumulated multiple open threads, unanchored decision verbs, or surfaced contradictions that would benefit from explicit graph tracking. Requires the dpd-mcp-server MCP server to be registered with Claude Code.
EnglishDump the full DPD graph as text. Default tree format; pass --format=mermaid for mermaid diagram. Use for snapshots, audit, or copy-paste into docs.
EnglishManual edit of a DPD node or pool item. Wraps closenode / addnode(provenance='manual') / poolreject / pool unsuppress (= clear rejectedat). Use when user wants direct control beyond ambient mode.
EnglishSubmit DPD dogfood feedback as a GitHub issue. Collects system metadata only (version, OS, agent, Python version, optional stack trace) — does NOT send any session content (no node text, scope, label, Pool, rejected hypotheses, or sessionid). User free-text description is the primary payload. Graph context is opt-in via manual copy-paste only.
EnglishGenerate inferred nodes for the current DPD graph: missing decompositions, unstated assumptions, gap candidates. Each inferred node requires user opt-in. Often paired with /fcot for falsification.
EnglishRetrieve past closed/archived DPD subgraphs whose FTS5 index matches a query, then distill selected results into graph-candidate proposals. User-pull only — Claude must not auto-invoke. Requires the dpd-mcp-server MCP server.
EnglishShow current DPD session state: active roots, focus node, Pool items (active + rejected), session mode. Use when user asks "where are we?" or wants a snapshot.
EnglishExtract decided / closed items from the DPD graph and render as markdown summary. Use at session wrap-up or to produce spec material from a complete subgraph.
EnglishExternally verify a DPD edge that claims a logically-necessary implication (layer='necessary'), using a context-stripped prompt so an independent verifier judges the implication on its own merits rather than rubber-stamping the agent's framing. Records the verdict via recordedgeverification. Invoke when the user runs /dpd-verify-edge, or proactively suggest at a natural pause when listunverifiededges is non-empty. Requires the dpd-mcp-server MCP server.
EnglishImport an external prose/spec/graph document as a hypothetical archived DPD subgraph. Used for systematic gap analysis: import → /dpd-fill → /fcot pipeline.
Latin
AGENT SKILLS ATLASIncludes Dpd Feedback, Dpd Find Similar, Dpd Status, and other workflows.
Claude Code
Codex
Cursor
GitHub Copilot
Gemini CLISupported by the Atlas install prompt. Skill-specific requirements vary.
Pack stats
What’s included
Dpd
Bootstrap a DPD (Decompose-Propagate Decision) session under the ambient overlay paradigm — DPD is a passive graph overlay that observes ongoing conversation and proposes graph updates collaboratively. Invoke when the user runs /dpd, or proactively suggest when a conversation has accumulated multiple open threads, unanchored decision verbs, or surfaced contradictions that would benefit from explicit graph tracking. Requires the dpd-mcp-server MCP server to be registered with Claude Code.
Dpd Dump
Dump the full DPD graph as text. Default tree format; pass --format=mermaid for mermaid diagram. Use for snapshots, audit, or copy-paste into docs.
Dpd Edit
Manual edit of a DPD node or pool item. Wraps closenode / addnode(provenance='manual') / poolreject / pool unsuppress (= clear rejectedat). Use when user wants direct control beyond ambient mode.
Dpd Feedback
Submit DPD dogfood feedback as a GitHub issue. Collects system metadata only (version, OS, agent, Python version, optional stack trace) — does NOT send any session content (no node text, scope, label, Pool, rejected hypotheses, or sessionid). User free-text description is the primary payload. Graph context is opt-in via manual copy-paste only.
Dpd Fill
Generate inferred nodes for the current DPD graph: missing decompositions, unstated assumptions, gap candidates. Each inferred node requires user opt-in. Often paired with /fcot for falsification.
Dpd Find Similar
Retrieve past closed/archived DPD subgraphs whose FTS5 index matches a query, then distill selected results into graph-candidate proposals. User-pull only — Claude must not auto-invoke. Requires the dpd-mcp-server MCP server.
Dpd Status
Show current DPD session state: active roots, focus node, Pool items (active + rejected), session mode. Use when user asks "where are we?" or wants a snapshot.
Dpd Summary Markdown
Extract decided / closed items from the DPD graph and render as markdown summary. Use at session wrap-up or to produce spec material from a complete subgraph.
Dpd Verify Edge
Externally verify a DPD edge that claims a logically-necessary implication (layer='necessary'), using a context-stripped prompt so an independent verifier judges the implication on its own merits rather than rubber-stamping the agent's framing. Records the verdict via recordedgeverification. Invoke when the user runs /dpd-verify-edge, or proactively suggest at a natural pause when listunverifiededges is non-empty. Requires the dpd-mcp-server MCP server.
Dpd Import
Import an external prose/spec/graph document as a hypothetical archived DPD subgraph. Used for systematic gap analysis: import → /dpd-fill → /fcot pipeline.
Quick start
- Choose your skills. Select individual skills above, or choose the entire pack.
- Copy the install prompt. Use Copy install prompt above the list.
- Send it to your agent. Paste into your coding agent’s chat. Your agent reviews the source and guides the setup.
Try it with your agent
After installation, start with a request like:
Use the Dpd skill to help with my task.
Ask me for the context you need before getting started.Requirements & license
- A coding agent that can access your project files and install skills.
- Your agent checks each skill’s instructions, dependencies, and permissions during setup.
- Reported license for the skills shown: Apache-2.0. The author’s terms apply.