feat(scroll): P6.6 agent driving loop + lifecycle + milestone e2e (guard enforce live) - #71
Merged
Merged
Conversation
…ne e2e (guard enforces live)
…tion, not vitest kill
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Land the native attention-anchored agent's driving loop + lifecycle and the milestone e2e that PROVE the spatial guard enforces LIVE — first test with a live proposer AND a live camera, so it de-latents everything P6.2–6.5 built (before this, no test published a camera → the guard guarded nothing).
What
runAttentionAgent(peer, {intervalMs, actor?, config?, now?, onTick?})— drivetick()on a cadence untilstop(). Each tick fully settles before the next is scheduled (setTimeoutAFTER the await, notsetInterval) so slow propose round-trips can't pile up;tick()'s single-flight guard is the backstop.tick()—stateis read/written across the await, so an overlapping tick would last-write-wins clobber the observer fold (P6.5 carry-forward: agent non-laundering camera resolution (gating) + ObserverState + awarenessKnown producer #66 F-06). A racingtick()gets an idle no-op. Any throw from the pre-propose section fail-closes to an idle result —tick()never rejects, and the error stays observable viaonTick.Gate
simplifier → 4 parallel inspectors (one lens each) → adjudicator PASS. First pass BLOCKED on B1 (belt untested — the e2e proved only suspenders): fixed by tracking PROPOSED apart from COMMITTED and driving the WHOLE cold set, so a reader-blind belt is forced to propose band blocks and the assertion catches it. Teeth-proven: sabotage
resolveCameras → []fails the milestone at the belt assertion (agentMilestone.test.ts:144); revert → green. Also landed FN1–5 (runner leak net, deferred-fake reschedule-guard test, fail-closedtick(), cold-tied durability, F-06 skip pin) + C1 (race test) + C2 (P6.5 cold-target teeth).tsc -b --noEmit= 0. The 3 skips are honest P6 guard hardening (P6.3 carry-forward): awareness observer, reconnect-gap fail-closed, self-guard #63 pins — P6.3/P6.4 server awareness (F-04/F-05) + the new F-06 proposer self-exclusion.Carry-forward (non-blocking, no caller hits them)
stop()can't abort an in-flightpropose; a commit that lands afterstop()mutates but fires noonTick.Both are lifecycle-ownership decisions.
Closes #70.
🤖 Generated with Claude Code
https://claude.ai/code/session_01X7WHDHurYHVibsjwvevzFr