agentleFS
Sign inStart free
MicrosoftKnown publishermicrosoft/PilotSwarm/templates/builder-agents/skills/pilotswarm-duroxide-versioning/SKILL.md

pilotswarm-duroxide-versioning

Skill14 starsChanged 8 months agoMITView on GitHub

Use when changing PilotSwarm durable orchestration behavior, continue-as-new state, or replay-sensitive control flow.

We split this skill into the rules it gives an agent

7 rules. About 0 are common, 2 are distinctive, and 5 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈2

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 5

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Keep the canonical latest target in the shared version constant, and use that constant for all new starts and all …

    line 13
  2. Very specific:

    Treat the OrchestrationInput payload as a durable wire format.

    line 15
  3. Very specific:

    If version N can continueAsNew from state point X, version N+1 must behave correctly when resumed from X.

    line 18
  4. Very specific:

    If N+1 only works when resumed from a new point Y, that is a versioning bug unless the older path stays on the old handler.

    line 19
  5. Very specific:

    If replay behavior, yield order, timer semantics, KV usage, command routing, or continueAsNew state changes, treat it as a new …

    line 20
  6. Distinctive:

    Add tests for both:

    line 21
  7. Distinctive:

    When you add a new orchestration version, update the version-upgrade test suite too.

    line 24

≈ “Common” is an estimate: at least 1 in 5 of the 308,232 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
SKILL.md

PilotSwarm Duroxide Versioning

PilotSwarm durable orchestrations replay from history. Treat every orchestration change as a wire-compatibility problem, not just a code diff.

Rules

  1. Freeze old orchestration behavior in packages/sdk/src/orchestration_<version>.ts.
  2. Keep the canonical latest target in the shared version constant, and use that constant for all new starts and all continueAsNewVersioned(...) calls.
  3. Stamp the source orchestration version into the carried input so the target version can normalize or branch on legacy state if needed.
  4. Treat the OrchestrationInput payload as a durable wire format. Every new latest version must accept and normalize input snapshots from the oldest orchestration version that is still registered in the codebase.
  5. Backward compatibility is behavioral, not only syntactic:
    • Deserializing the old shape is not enough.
    • If version N can continueAsNew from state point X, version N+1 must behave correctly when resumed from X.
    • If N+1 only works when resumed from a new point Y, that is a versioning bug unless the older path stays on the old handler.
  6. If replay behavior, yield order, timer semantics, KV usage, command routing, or continueAsNew state changes, treat it as a new orchestration version even when the diff is small.
  7. Add tests for both:
    • the version target (continueAsNew upgrades to latest as intended)
    • the compatibility path (older carried input still resumes correctly)
  8. When you add a new orchestration version, update the version-upgrade test suite too. In particular, keep packages/sdk/test/local/orchestration-version-upgrade.test.js aligned so it still exercises upgrade from the last three frozen versions into the latest version.

Checklist

  • Shared latest version constant updated
  • New latest handler added and registry wired
  • Frozen handlers still replay-safe
  • Oldest registered input shape still normalized by latest
  • Source orchestration version carried in input
  • Local tests cover at least one frozen-to-latest handoff
  • packages/sdk/test/local/orchestration-version-upgrade.test.js updated for the new latest version and the rolling last-three compatibility matrix

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

No reviews yet

People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.

  • It worked
  • Partly
  • It didn't
What did you use it for? e.g. Reviewing pull requests in a Rails app

Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free