2026-08-29 - 2026-09-29
Overview
76 Pull requests merged by 4 users
Merged
#147 fix(alignment): resolve /root/ hardcoding and stale tanko-DSH references
Merged
#145 fix(alignment): clarify tanko live LiteLLM proxy status in health-check description
Merged
#144 fix(alignment): use llmuser with sudo for swap-gpu-dense-model.sh
Merged
#141 fix(audit-hermes): handle fallback_providers as list or dict
Merged
#140 fix(hermes): clarify auxiliary model policy to match audit
Merged
#139 feat(memory-fixer): add duplicate-detection phase; correct stale canonical pointer + reporting model (v2.2.0)
Merged
#138 feat(search): agent-consumption layer — dedupe, filter, rerank, extract content
Merged
#137 feat(daily-digest): deliver via Zulip DM as an HTML attachment; drop mail entirely
Merged
#136 fix: reconcile the health-logs self-heal contracts, and add a dead-man's-switch
Merged
#134 feat: land the revision-preflight guard, fixed and wired into contract execution
Merged
#135 feat: add the missing daily-health-digest contract + fix the red lint from #133
Merged
#133 fix(daily-infra-report): resolve PVE token from env; make a dead probe non-silent
Merged
#132 feat: multi-engine search stack + visibility check
Merged
#131 Fix PBS GC monitor false positive and add contract-run.sh wrapper
Merged
#129 feat(security): commit-time secret guard that FAILS the build on a committed credential
Merged
#127 fix(daily-infra-report): drop the vestigial Zulip key requirement; fail loudly on a failed send
Merged
#126 fix(daily-infra-report): missing credentials degrade one leg instead of blacking out the digest
Merged
#125 feat(litellm): align contracts for 7-provider cloud consolidation (2026-09-20)
Merged
#124 fix(zulip-monitor): add C3 public access path, make run verdict non-optimistic, separate C1/C2/C3
Merged
#122 fix: Rule 5 accept canonical internal and public host base_url
Merged
#123 fix(litellm-health): three-state model verdicts - busy is not a failure
Merged
#121 fix(litellm-health): retry single-host model probes once at a longer timeout
Merged
#120 fix(infra-monitoring): probe the real Docker Stats and PVE exporter ports
Merged
#119 feat(proxmox-monitor): add PBS GC liveness signal
Merged
#115 fix: move infra-monitoring probes into versioned script
Merged
#118 fix: update MCP access docs to reflect per-key grants support
Merged
#114 feat: add MCP server URL validation to hermes-config-template contract
Merged
#112 security(secrets): remove committed credentials from the tree and read them from the vault/environment
Merged
#111 docs(agent-health): every check leg must appear in every report - a missing line is not a pass
Merged
#109 fix(monitoring): make the reported key count self-describing instead of a bare number
Merged
#108 docs(keys): state the acceptable key-placement pattern and the backup-file fix procedure
Merged
#107 docs(contracts): add the missing agent-health-check contract
Merged
#106 fix(disk-gc): actually write the host-band state file so escalations and recoveries can fire
Merged
#105 fix(disk-gc): host filesystem bands, named volumes, report-only, and state-change alerts
Merged
#104 docs: pm2/spoton AS-BUILT correction and the real Prometheus node coverage
Merged
#77 Remove client deliverable from the contracts repo (process fix)
Merged
#103 ci: make PR validation unconditional - the paths filter silently skipped whole classes of pull requests
Merged
#97 fix(backups): document the thin-pool headroom and staging-directory preconditions, with the incidents that justify them
Merged
#95 fix(keys): correct the key-lifecycle contracts to the measured truth
Merged
#93 fix(agent-health): gateway liveness reports a probe failure as a probe failure, never as an agent outage
Merged
#92 fix(monitoring): probe precision - any HTTP status means alive, failed probes never become service verdicts
Merged
#91 fix(disk-gc): deterministic reachability verdict and correctly labelled disk figures
Merged
#90 fix(hermes): separate POLICY observations from FAULT findings in the audit contracts
Merged
#89 fix(hermes): reachability verdict must not come from the remote command's exit code
Merged
#88 fix(monitoring): litellm-health script - pool-alias timeout and leaked key-list debug output
Merged
#87 feat(monitoring): add scripts/litellm-health-check.py so the litellm-health contract is executed, not improvised
Merged
#86 fix(contracts): abiba default model row + disk-gc access-method note
Merged
#84 fix(litellm-health): run the key-list admin call on the gateway host, not inside the container
Merged
#83 fix(monitoring): make credential sourcing explicit and fail loudly when missing
Merged
#82 fix: restore per-host probe coverage + sweep residual retired names
Merged
#81 fix(disk-gc): hard guest-level report-only gate for CT 111/.129 + correct stale fleet map
Merged
#80 fix(audit): stop requiring the retired gpu-light alias; derive model fields; sweep retired names
Merged
#79 docs(litellm-health): single source of truth, gpu-vision alias, litellm-health as live owner
Merged
#78 fix(zulip-health): retire the dead kagentz Zulip adapter leg
Merged
#76 fix: correct Agent Zero A2A probe port in zulip-monitor.sh
Merged
#75 feat(litellm): 1.99.1 update coverage, trove agent, nginx /ui /docs fixes
Merged
#74 fix(router): decommission legacy GPU router across contracts and fleet scripts
Merged
#73 fix(dsh-web-auth): Authentik-gated :80 login + non-disruptive token capture
Merged
#72 feat(memory-fixer): v2.1.0 — stale nodes auto-archived (Kwame directive 2026-09-11)
Merged
#64 update(infrastructure-update): v1.3.0 — full 5-host Docker ecosystem coverage (live-verified 2026-09-08)
Merged
#71 fix(monitoring): stop monitoring Mumuni from this host (moved to CT105/.14)
Merged
#70 fix(monitoring): probe-drift round 2 — gpu port, PVE liveness, report-only visibility, report provenance
Merged
#69 fix(zulip-monitor): read nested zulip.connected; probe failures alert and never restart
Merged
#68 fix(zulip-monitor): stream alert body carries real content; WARN on delivery failure
Merged
#65 fix(agent-health-check): correct unit names, abiba key path, pid crash, uptime math (script-only)
Merged
#66 fix(zulip-health): tanko probe via amdpve pct + loopback-any-HTTP alive rule + zulip-monitor.sh stale-path fix
Merged
#67 provision: register okyeame-memory-audit cron job id
Merged
#63 fix: probe alignment across multiple monitoring contracts
Merged
#61 feat: litellm-client-timeouts — standard client timeout/retry policy for all agents
Merged
#60 fix: restore agent-health-check AGENTS roster + land missed Tanko-DSH rows
Merged
#59 fix: remove duplicated ## Execution block in infrastructure-monitoring
Merged
#58 fix: rename agent-zero fix-summary out of contract scan path
Merged
#57 Fix zulip-monitor.sh: Remove email spam, use Zulip #agent-hub only
Merged
#56 fix: repoint Mumuni refs in Normal-sensitivity contracts to kagentz CT105/.14
Merged
#55 fix: correct Mumuni gateway lifecycle path (no sudo on kagentz) + 2 leftover CT100 mentions
Merged
#54 fix: repoint Mumuni refs from CT100/.24 to kagentz CT105/.14 post-migration
2 Issues created by 1 user
Opened
#110 litellm-health key count: two follow-up notes from the review (non-blocking)
Opened
#113 Follow-ups from the credential purge (#112): refusal diagnostics, Key Prefix labels, truncated prefixes in rotation history