feat: multi-agent macroeconomics data pipeline example#6
Open
feat: multi-agent macroeconomics data pipeline example#6
Conversation
Orchestrator-miner-modeler pipeline that downloads World Bank WDI data, extracts indicators via agent-xlsx, and builds a formatted Dashboard.xlsx. Includes sandboxed execution via @anthropic-ai/sandbox-runtime.
Add a reusable skills provider module (src/lib/skills/) that bridges the agentskills.io ecosystem with zeitlich agents via InMemoryFs lazy files. Skills are fetched from GitHub on first read and cached thereafter. Integrate the agent-xlsx skill into all three macro-agent workflows (orchestrator, miner, modeler) so agents discover CLI commands, flags, and patterns from the skill reference rather than hardcoded prompts. Key changes: - src/lib/skills/: types, catalog builder, lazy mount + load-skill command - Macro-agent prompts de-hardcoded: agents derive titles, years, columns, and layout from the data rather than from prescriptive system prompts - All sandbox/ path prefixes removed from prompts (tool runs from sandbox) - Tool description trimmed to structural info (skill owns command reference) - Remove redundant Dashboard.xlsx copy-to-root from start.ts
Add comprehensive documentation covering architecture, sandbox design, skills provider, and key design decisions. Add assertAgentXlsxInstalled preflight in the worker to fail fast if the CLI is missing from PATH.
Contributor
|
@apetta, we could try to update this to the latest zeitlich version, lots of the features needed are now built in. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary
New
macro-agentexample — a multi-agent pipeline that downloads the World Bank WDI dataset (~76MB, 401K rows), extracts economic indicators via specialized subagents, and builds a formatted Excel dashboard.agent-xlsxCLI calls via@anthropic-ai/sandbox-runtime(Apple Seatbelt on macOS, bubblewrap on Linux)src/lib/skills/) that lazy-loads agent-xlsx CLI docs from GitHub intoInMemoryFs, so agents discover commands from skill references instead of hardcoded promptsAgentXlsxpassthrough tool rather than per-subcommand typed tools, avoiding schema drift as the CLI evolvesagent-xlsxis not on PATH