Index of every page on this site: llms.txt
All skills
av avatar

/workgraph

@8f70341
by Ivan Charapanauav/skills28 stars
2
Run it yourself

Orchestrate work as a cyclic directed graph of subagent-executed nodes with transition criteria on edges, budgeted cycles, and per-node gates (runtime verification, metric, or artifact). Use when the user says "workgraph", "run this as a graph", "graph this work", or when a goal has parallel branches, feedback loops, or mixed gate types that a linear loop cannot express. Every node is sized to a 64k context budget, so small goals run as a two-node do/verify graph and large goals scale out without caps. For clock-driven or metric-driven single loops, timeboxed-iterating and autoresearch remain lighter. Parent orchestrators load this file; a one-node worker follows its node prompt and input file instead, unless its charter assigns planning, routing, gating or graph maintenance.

Use this Skill: https://skilld.dev/gh/av/skills/workgraph

Nothing lands on disk. Nothing to clean up.

Fork this Skill

Edit a local copy. It keeps the author and licence.

README.extra.md

≈641 tokens on demand. Your agent reads this file only when SKILL.md points to it.

Example graphs

Every render uses the skill's fixed palette: dashed PENDING · outlined READY · bold RUNNING · gray VERIFIED · red FAILED · amber ESCALATED

Database migration with dual-write and a rollback gate

dual_write and backfill run in parallel and rejoin at the parity gate. Parity found row drift, so the FAIL edge sent backfill back for a second pass (×2). A failed soak routes to rollback, and from there back to plan.

<picture> <source media="(prefers-color-scheme: dark)" srcset="./assets/graph-db-migration-dark.png"> <img src="./assets/graph-db-migration-light.png" width="860" alt="Database migration graph: plan, schema and dual_write verified; backfill ×2 running after parity failed; read_switch, soak, rollback, drop_old and cleanup pending"> </picture><picture> <source media="(prefers-color-scheme: dark)" srcset="./assets/graph-legend-dark.png"> <img src="./assets/graph-legend-light.png" width="735" alt="Key: hollow ring PENDING, solid dot READY, ringed dot and bold label RUNNING, check VERIFIED, red dot FAILED, amber triangle ESCALATED; red dotted line = FAIL edge; gray dotted line = loop back; ×N = visit count"> </picture>

Multi-source research report with fact-check and revision

scope fans out to four source nodes that rejoin at synth. forums escalated (the source was unreachable) and the report went ahead without it. fact_check rejected the first draft, revise fixed it, and the second check is running now. A critic FAIL goes back to synth.

<picture> <source media="(prefers-color-scheme: dark)" srcset="./assets/graph-research-report-dark.png"> <img src="./assets/graph-research-report-light.png" width="640" alt="Research report graph: scope, papers, filings, news, synth and revise verified; forums escalated; fact_check ×2 running; critic and publish pending"> </picture>

Release with a test matrix, canary and rollback

build fans out to a four-way test matrix behind a single gate. The first canary tripped on a 5xx spike, so rollback and hotfix ran and the graph looped back into build. Second visits (×2) are in progress on macOS and e2e, while notes waits at rollout.

<picture> <source media="(prefers-color-scheme: dark)" srcset="./assets/graph-release-canary-dark.png"> <img src="./assets/graph-release-canary-light.png" width="730" alt="Release graph: freeze, build ×2, linux ×2, windows ×2, rollback, hotfix and notes verified; macos ×2 and e2e ×2 running; canary failed; gate and rollout pending"> </picture>

Source: SKILL.md on GitHub

skilld matched fixed text patterns in SKILL.md and file names. Patterns miss obfuscated code.

skilld run checks every file with the same patterns. It asks for approval before it loads a Skill with a behavior marked Needs approval.

No alerts8d3 checks · Risk SAFE
  • Gen Agent Trust Hub9d

    The skill orchestrates work using a graph of subagents. It is susceptible to indirect prompt injection because the orchestrator executes 'verification' shell commands generated by subagents, which could be manipulated by malicious data in the target repository. It also performs standard repository operations like git worktree management and merging.

  • Socket8d

    No alerts

  • Snyk9d

    Risk: LOW · No issues

Signed by skilld at 8f70341. This ties the file your Agent reads to that commit on GitHub. It does not review the instructions.

Last checked against GitHub last week.

Activeupdated last week

README badge

README badge for av/skills/workgraph