Repository navigation
Add a shared Python supervisory-control example - #72
Merged
Merged
Conversation
Pitot coding-agent E2EIntelligence Flow is the verification source; Pitot's public README carries the latest
Runtime capabilities
Source commit: |
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
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Change contract
When: A developer wants Codex to delegate to Qwen or DeepSeek while a shared application controls which delivery actions are admissible.
Before: The examples show individual approval decisions, without a stateful delivery supervisor or worker recovery from a denied action.
After:
examples/supervisory-controlruns real agent CLIs through one Pitot Controller and a shared Python supervisor. A local model replay drives both worker variants through early release denial, completed build, external test invalidation, recovery and verified release. The README explains the control concepts, practical use cases, program-first supervision, replacement policy boundaries and a sourced MCAS engineering lesson; technical comments describe the code flow.Boundaries preserved
Evidence
ea9773e4a9f4Failure and recovery
Unknown commands and wrong roles are denied. An invalidated or changed artifact must be rebuilt and tested before release. The example cleans up owned processes and temporary files; optional logs preserve native feedback for diagnosis. This is an additive example and can be reverted without changing the library API.
Remaining limits
The runnable native demo targets macOS and Linux and runs one delivery session with one worker at a time. It demonstrates control boundaries using deterministic local model replay; it does not measure model quality, provide a hostile-process sandbox, or implement post-release revocation. The README lists tested CLI versions and reproducible commands.