Captures uncommitted work present on the live droplet so origin/main can be integrated and the new control toolbar deployed. Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Vision reports a display name ("Karti") while the git watcher reports a handle ("karti"); the detectors compared engineerId case-sensitively, so the same human collided with themselves. Canonicalize identity at the decision point in both the file and research detectors, keeping the display-cased name for the card.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Voice was gated to severity === 'critical', so most conflict cards were
silent. Now every intervention is spoken: priority is derived from
severity in publishHermesIntervention — critical jumps the queue,
the rest serialize so concurrent alerts don't garble each other.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LuV8W8oNYRsDWKoqK8Mkqc
The edge-trigger single-shot keyed conflicts by file basename only and
re-armed solely when the conflict left a detection cycle. README is dirty
for someone in the pod the whole session, so its key never re-armed: it
alerted once (ram+karti at restart) and stayed muted even as the colliding
pair changed to karti+Ramis Hasanli. Transient files (research, new files)
churned out and back, so only those kept firing.
- conflictKey now includes the sorted engineer pair, so a different pair on
the same file re-alerts immediately.
- activeConflicts is now Map<key, lastAlertedAt>; a persistent conflict
re-alerts every CONFLICT_REALERT_MS (default 20s) instead of going silent
forever, without spamming once-per-frame.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LuV8W8oNYRsDWKoqK8Mkqc
Four demo-safety fixes on top of #43, none touching demo-pod data:
- De-dup at source: the suppression path returned before activeConflicts.add(),
so the same unresolved dismissed collision wrote a SuppressionDoc every frame
(~194 spam rows observed on prod). Now mark the conflict handled first, so it
records ONCE per recurrence and re-arms via the onScreenFrame resolution sweep.
- Await the write: recordSuppression is the visible learning proof, so await it
(like recordCollision/recordIntervention) instead of void ...catch().
- Display de-dup: collapse suppressed beats to one per file (most recent) with a
stable per-file id, so any pre-fix duplicate rows never render as spam.
- Live-graph gate: suppression beats now satisfy the "has activity" check in
materializePodGraph, so a clean pod with preserved suppressions (but no
collision/file nodes) no longer falls back to the demo graph and hides the proof.
- Ops: /api/memory/stats counts `suppressions`; docs/mongodb.md documents the
collection and flags it "preserve in DB cleanup" (visible learning evidence).
Verified end-to-end on a throwaway pod (not demo-pod): suppression-only pod
materializes; 2 dupe rows render as 1 beat. backend+frontend typecheck + eslint pass.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Dismissing one collision permanently silenced all future ones: the
dismissal gate (priorOutcome.accepted === false) plus loose file/vector
recall meant one README discard muted every later README clash. The
per-pod cooldown additionally hid consecutive conflicts for 3 min.
shouldIntervene now only filters `info` severity — Accept/Dismiss still
record outcomes but no longer gate whether a collision surfaces. The
same-collision single-shot dedupe (activeConflicts in PodmanAgent.handle)
still prevents repeat spam, so removing the cooldown is safe.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LuV8W8oNYRsDWKoqK8Mkqc
Codex review fix. The earlier approach synthesized a 'suppressed' beat from every
dismissed OUTCOME, which is wrong: a dismissal is not a later suppressed repeat,
and the old dismissal timestamp sank below the 12-row activity cap (invisible).
Now the negative-feedback loop is recorded WHEN IT HAPPENS. When shouldIntervene()
returns false specifically because a signature was dismissed before
(agent/podman.ts), the agent writes a durable SuppressionDoc to a new
`suppressions` collection, timestamped at the repeat. graph/live.ts materializes
those into kind:'suppressed' activity (recent -> surfaces at the top). The
suppressed collision is never written to `collisions` (agent returns before
recordCollision), so this is its own record.
Verified end-to-end: a transient demo-pod suppression renders at the TOP of the
activity stream; cleaned up after. shared build + backend/frontend typecheck +
eslint pass.
Co-authored-by: Claude Opus 4.8 <noreply@anthropic.com>
Surfaces "rework saved" at the top of the teammate Work History dialog:
a transparent heuristic over collisions Hermes caught early (eligible =
gitOverlap or critical, with an intervention), credit split across
involved engineers. Shows hard counts (clashes caught, conflict-free
files) plus an info-tooltip breakdown of the estimate.
- shared: optional MemberWorkHistoryRoi field (back-compat, self-zeroes)
- backend: query collisions + interventions in getMemberWorkHistory,
computeRoi helper
- frontend: RoiBand + RoiTooltip components, hidden when no clashes
Implements docs/plans/work-history-roi.md.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01LuV8W8oNYRsDWKoqK8Mkqc
Addresses the P1 brittleness in the Step 3 verifier so learned_from edges
(graph/live.ts:413) can't be silently zeroed on stage.
- Capture overlap AT detection time as Collision.gitOverlap (podman.ts), while
engineer_states are still fresh, instead of re-deriving from possibly-stale
state when the user clicks. deriveWasRealCollision() now prefers this stored
evidence and only falls back to a live re-derivation for older collisions.
- Canonicalize engineer names (trim + lowercase) on both the capture and the
fallback path, so "Karti" vs "karti" no longer misses the git state.
- conflictKey now reuses the shared comparableBasename() helper (dedupe).
shared/src/collision.ts gains optional `gitOverlap?: boolean` (additive).
shared rebuilt; backend+frontend typecheck + eslint pass. PLAN.md rung 3 updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Backend becomes authoritative for the verifier signal. recordOutcome now
overrides the client-supplied wasRealCollision with deriveWasRealCollision():
a flagged collision counts as REAL only if BOTH named engineers currently have
the collided file in their git changedFiles (getGitStates, 120s freshness TTL).
Conservative false when the collision is orphaned/missing or git state is
stale. Restores the (accepted x wasReal) 2x2 the spec assumes instead of the
hardcoded 107/107 true.
frontend/useInterventions.ts stops sending a hardcoded `true` (now a
backend-overridden placeholder). Spec: continual-learning/spec.md:98-108,
policy.md:35-42. backend+frontend typecheck + eslint pass. PLAN.md P0.5 rung 3
added.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Step 1 — policy.ts shouldIntervene: suppress on a prior dismissal alone.
The former `&& !priorOutcome.wasRealCollision` term was dead code (outcomes
record wasRealCollision hardcoded true), so the 85 real dismissals in Atlas
were never used. Now a prior accepted===false suppresses the next identical
nudge. Spec: continual-learning/policy.md:41, spec.md:163.
Step 2 — podman.ts handle: only escalate severity to 'critical' (the spoken
alert trigger) when the recalled prior was an accepted *real* collision,
instead of blanket-escalating every recall. Surfaces learned routing in
preferredAction; stops dismissed/false priors over-escalating to voice.
Spec: continual-learning/policy.md:62-63, plan.md:66.
Documentation-first: adds PLAN.md section 8 "P0.5 - RSI negative-feedback
activation" with both rungs + follow-ups. No schema change. backend typecheck
passes. Independent of the MongoDB-cleanup handoff (Codex).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The 'Test audio' button becomes 'Background Music': each pod gets a calm looping track from Gemini Lyria 3 that sings the pod name once up front then stays instrumental. Backend GET /api/pods/:id/music generates via the Gemini interactions endpoint and caches the MP3 per pod in Mongo (pod_music); the frontend fetches and loops it via Web Audio, published pod-wide on the existing podman-beat track.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Lands the dynamic Team-memory redesign on top of main's continual-learning
pipeline. main already computes loop/activity in the API but its GraphView never
rendered them and kept a static-column graph; this swaps in the dynamic graph and
surfaces the rails, reusing main's richer PodLearningLoop / PodGraphActivity types.
Frontend (new frontend/src/components/graph/*, composed into GraphView.tsx):
- forceSim.ts: dependency-free force layout (charge, link springs, centroid
recenter, 2-pass collision, bounds clamp, alpha anneal) — no new deps/lockfile churn.
- GraphCanvas.tsx: SVG render from the sim — draggable + pinnable nodes, curved
edges, weight-sized shapes, fade-in, animated learned_from dash, risk-path
lighting / rest dimmed, label collision-avoidance.
- MetricsRail / LearningLoop / ActivityStream / SelectedNodePanel / encoding.ts:
the mock's rails + stream + detail panel in light shadcn. LearningLoop consumes
main's PodLearningLoop (steps + activeStep); ActivityStream consumes
PodGraphActivity (title + detail). SelectedNodePanel adds a Flow section that
narrates the path through a clicked node (flowNarrative); default copy is
mode-aware. Edge legend rounded out (editing/touches).
- GraphView polls every 5s and diffs (positions preserved), + ws /api/events nudge.
Stale selection (node gone across a poll) dropped so the canvas can't dim entirely.
Backend (surgical — main's materializer + buildLoop kept):
- live.ts: headline metric cards derived from the FINAL de-noised graph (Open risk
paths = distinct collision files; Learned owners = distinct owner engineers)
instead of raw collision-signature / accepted-outcome counts that inflate with
test churn (50 -> 4 risk files, 16 -> 1 owner on live). buildLoop untouched.
- demo.ts: metrics realigned to the demo graph (3 owners / 1 risk path / 100%).
Verified: lint + -r typecheck + -r build pass; Playwright confirmed the dynamic
graph (ticks on load, draggable), the loop rail (5 steps, ADAPT active) and
activity stream rendering main's shapes, honest metrics (3/1/100%), and the flow
narrative per node kind — zero page errors.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Both dedup gates (shouldIntervene per-pod cooldown + hasRecentInterventionForCollision per-file window) were purely time-based, so an unresolved conflict re-alerted every NUDGE_COOLDOWN_MS (30s on prod). Track active conflicts in-memory in PodMan and alert only on the transition into conflict; re-arm when a detection cycle no longer sees it.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
The deployed graph was an unreadable mess: full file paths centered on nodes
bled across columns, and collision labels showed garbage vision symbols
("infra/README.md### Running the git watcher").
- node labels are now the last two path segments (full path in summary)
- collision labels are just the file (drop the junk #symbol)
- cap file nodes to 9 (collision files prioritized)
demo-pod: 22 -> 20 clean nodes; no label bleed.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Materializer (live.ts) — cut demo-pod from 110 -> 22 nodes:
- cap to 8 recent collisions; collapse repeats by memorySignature
- collapse interventions to one (most recent) per collision
- filter junk 'files' (URLs, env vars, browser/app names, scratch/test) to
real source paths only
- prune test-artifact engineers (a/b/verify/codex-check/-testrepo) + any node
orphaned by that
- collisions referenced by accepted outcomes bypass the cap so the learned_from
money path never drops; risk-paths metric counts distinct signatures
UI: per-pod 'Team memory' action on each PodCard's menu (replaces the header
button that always opened pods[0]).
Verified against live demo-pod data (joins connect); typechecks clean.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Voice generation, critical escalation, and audio publish all worked, but
participants heard nothing: browsers block autoplay of the agent's audio
track until a user gesture. Add the canonical LiveKit unlock — listen for
AudioPlaybackStatusChanged and show an "Enable sound" button that calls
room.startAudio() so PodMan's voice cues are actually heard.
Also make collision alerts short and direct instead of chatty AI prose:
card: "Conflict: ram + yahya both on README.md (unpushed). Seen before."
voice: "Conflict. ram + yahya, both on README.md."
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AaCFWMkYQmTcuPsxaaACft
Verified materializePodGraph against real demo-pod data (joins all connect:
21/21 interventions->collisions, learned_from produced). Tuning:
- engineer nodes are case-insensitive (merges Shakthi/shakthi)
- git (engineer_states) now only CONFIRMS editing on files vision/collisions
already surfaced, instead of adding the whole repo diff (killed a 29-file
node explosion from a watcher running against the full podman repo)
Cut demo-pod from 110 -> 77 nodes; git file-edges 39 -> 6.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- backend/src/graph/live.ts: materializePodGraph builds the graph from the real
collections (pods, engineer_states, observations, collisions, interventions,
outcomes) instead of the demo seed. Parses git-status paths, fuses git+vision,
draws collides/warns/learned_from, computes live metrics + a column layout.
- store.ts: loadPodGraph now prefers live → seeded team_model.graph → demo.
- GraphView.tsx: light/shadcn theme (from the team-memory-theme work), composed
from the ruixen primitives; node-shape encoding unchanged.
- docs/graph.md: live data→graph mapping + fallback order.
Typechecks clean. Not yet runtime-verified end-to-end (Atlas creds rotated again).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Live observations showed two engineers both dirty on README.md but zero
collisions firing. Two root causes:
1. normalize() only lowercased and prepended src/, so the same file at
different path depths ("agent.ts" vs "backend/src/agent.ts") never
matched. Replace with lowercased-basename matching.
2. Git ground truth (engineer_states.changedFiles) was only used to set
the unpushed flag, never to match the file. The strongest signal was
wasted.
Detector now fuses vision currentFile AND git changedFiles into one
file->engineers map, so a collision fires when two people share a dirty
file even if both screens aren't on it at the same instant.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AaCFWMkYQmTcuPsxaaACft
Reverse the best-effort error swallowing. MongoDB is core to PodMan's
continual-learning story and must always be used, so a broken memory
layer must surface immediately rather than silently masquerade as
working (which is how observations stayed at 0 unnoticed).
- agent verifies Mongo via initMemory() at boot; bad creds / unreachable
Atlas now fail loudly before joining the room, not mid-demo
- server exits on Mongo init failure instead of warning and limping on
- getGitStates and onScreenFrame no longer swallow Mongo errors
- memory persist() logs the failure and rethrows instead of warning
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AaCFWMkYQmTcuPsxaaACft
A Mongo auth/connection failure in getGitStates escaped uncaught and
killed the agent process on the first screen frame, so collision
detection never ran. Make git-state fusion best-effort (degrade to
vision-only) and wrap the whole onScreenFrame loop so no per-frame
Gemini/GitHub/Mongo error can crash the long-running agent.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AaCFWMkYQmTcuPsxaaACft
Resolve App.tsx header conflict: keep main's "Privacy-limited" badge AND the
"Team memory" graph button (both in the header action group). server.ts and
backend/package.json auto-merged cleanly — main's closeMemory/shutdown and the
graph:seed script are both preserved.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Resolve conflicts:
- backend/src/server.ts: keep main's getPresence + closeRoom (room auto-cleanup)
and add the graph imports/routes (additive).
- frontend/src/App.tsx: take main's shadcn command-center UI and re-integrate
GraphView — import, graphPodId state, an early-return render branch, and a
shadcn "Team memory" button next to Refresh.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
- reachFrom now returns the direct outbound edges PLUS the recursive
$graphLookup reaches, de-duped by edge id, with maxDepth to bound cycles.
It was dropping edges straight off the start node. [review P2]
- GraphView SVG nodes are keyboard-accessible: role/tabIndex/aria-label +
Enter/Space handler alongside onClick. [review P3]
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Adds the team_model graph layer (the "it learned" view) and its visualization,
per docs/graph.md. Demo-backed first; built on origin/main conventions.
- shared: PodGraph / node / edge / metric types + GraphNodeDoc / GraphEdgeDoc
- backend/src/graph: demo graph data; Mongo store (loadPodGraph w/ demo
fallback, seedGraph, reachFrom via $graphLookup); graph:seed script
- backend/src/server.ts: additive GET /api/pods/:id/graph + /graph/reach/:node
- frontend: dark-Bauhaus GraphView + fetch helper, reachable from a
"Team memory" toggle in App.tsx
- docs/graph.md: spec (data model, $graphLookup, API, demo-first plan)
Demo-backed: the route serves a grounded demo graph when team_model has no
graph yet; graph:seed writes team_model + graph_nodes/graph_edges so the
$graphLookup traversal is real. Swap loadPodGraph to live team_model later.
Known follow-ups before merge: branch is based on 65a0791; origin/main is now
at 1294b84, so this needs a rebase + App.tsx conflict resolution (teammate
rewrote App.tsx with live room view / one-click join).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
DELETE /api/pods/:id now also calls closeRoom (RoomServiceClient.deleteRoom)
to end the live LiveKit room and disconnect anyone connected. PodView listens
for RoomEvent.Disconnected and returns the user to the pod list, so a deleted
pod cleanly kicks everyone out instead of stranding them in a dead room.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Use the member name as the LiveKit identity (was random per join) so a
refresh/rejoin replaces the existing session instead of leaving a ghost
participant in the room. Set roomConfig on the token (emptyTimeout 60s,
departureTimeout 20s) so empty rooms and departed participants clean up
promptly. Pods map 1:1 to a deterministic room name, so joins reuse the
same room rather than spawning new ones.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
Add getGitStates(podId) to db.ts — queries the engineer_states collection
written by scripts/podman-agent.mjs and returns a map keyed by engineer name.
In PodMan.onScreenFrame(), fuse before detectCollisions(): if an engineer has
changedFiles in engineer_states, force hasUnpushedChanges=true on their context.
This overrides Gemini vision's unreliable pixel-based guess with deterministic
local git state — preventing the money-moment collision from silently not firing.
Key assumption: LiveKit participant identity matches the --name arg used when
running the git watcher (e.g. identity "alice" → --name alice).
Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01AaCFWMkYQmTcuPsxaaACft
Yahya Alhinai