[uk ai resilience] UK AI Open Code Risk & Resilience Governance — Weekly Report 2026-07-01

[github-actions[bot]]

Executive Summary

Weekly UK AI Open Code Risk & Resilience Governance scan for github/gh-aw, lookback 2026-06-24 → 2026-07-01. 451 commits merged, 110 flagged for security signals, 15 open security-labelled issues, 8 open code scanning alerts, 0 secret scanning alerts.

Overall posture: Tier B (Open With Conditions) — one active Tier C finding (AIC guardrail starvation, tracked as #42761), five Tier B findings requiring time-boxed remediation. No Tier D findings. Repository is actively maintained with strong CI/CodeQL coverage; primary gaps are in alert triage velocity and cross-run operational coordination.

---

Asset Graph (recent-change scoped)

Asset	Area	Changed	Alerts	Issues	Tier
AIC Guardrail / Rate-Limit	actions/guardrail	✅	—	#42761	C
Parser (schedule_fuzzy_scatter)	compiler/parser	✅	#637 HIGH	—	B
Project CLI (GraphQL)	cli	✅	#627 #628	—	B
Publish pipeline (CI action)	actions/deps	—	#625 MED	—	B
Scripts (docs/evals)	tooling/scripts	—	#635 #636	—	B
Engine/Skill Auth surface	auth/engine	✅	—	#42749	B

---

Tier Classification Table

Area	Tier	Rationale
AIC Guardrail	C — Restricted Pending Review	Real incident: 99.5% GITHUB_TOKEN budget consumed, cross-workflow starvation proven. On by default every run. No cross-run coordination.
Parser Integer Overflow	B — Open With Conditions	CodeQL HIGH (CWE-190/681), strconv.Atoi→uint32 no bounds check. Trivial fix. Fresh alert (2026-07-01).
GraphQL Injection (CLI)	B — Open With Conditions	CWE-89, two unescaped node ID injections (15 days unaddressed). Low direct exploitability but defence-in-depth gap.
Unpinned Supply Chain Action	B — Open With Conditions	docker/build-push-action@v7.2.0 tag in publish-safe-outputs-node.yml (17 days open). Supply chain vector for image compromise.
Script HTTP-to-file / Insecure Temp	B — Open With Conditions	CWE-434 and CWE-377/378 in tooling scripts. Not core runtime; medium urgency.
Engine/Skill Auth	B — Open With Conditions	Expanding credential surface (per-skill auth, engine agent type). Misleading leak warning (#42749) may reduce operator trust.

---

Control Verification Gaps

Domain	Status	Gaps
Ownership	Partial	#637, #627/#628, #625 unassigned; #42761 has no assignee
SDLC	Partial	Alert triage SLA undefined; #627/#628 open 15 days without response
Dependency	Partial	#625 unpinned action 17 days open; no global action-pinning policy
Secret Exposure	Good	0 secret scanning alerts open
Runtime Observability	Partial	No cross-run/aggregate guardrail cost metric; no budget-threshold alert
Recovery	Partial	MTTR ~7–15 days for HIGH alerts; #42761 mitigations not yet implemented

---

Risk-Scoring Table

Area	Exposure	Patchability	Detectability	Fragility	Ownership	Aggregate	Tier	SLA
AIC Guardrail	5	3	3	5	3	19	C	Critical (<24h)
Parser Int Overflow	2	1	4	2	4	13	B	High (<7d)
GraphQL Injection	2	1	4	2	3	12	B	High (<7d)
Unpinned Action	3	1	4	4	3	15	B	High (<7d)
Script HTTP/Temp	2	2	4	2	3	13	B	Medium (<30d)
Engine/Skill Auth	3	2	3	3	4	15	B	Medium (<30d)

---

Remediation Queue

Critical (<24h): #42761 AIC Guardrail — implement O(1) aggregate-cache read; widen fallback; add in-loop rate-limit backoff; decide on isolated token.

High (<7d): CodeQL #637 parser int overflow; Semgrep #627/#628 GraphQL injection; CodeQL #625 unpinned action.

Medium (<30d): Script file ops (#635/#636); engine/skill auth audit (#42749).

Operational Metrics

Metric	Value	Target
MTTR proxy (HIGH alerts)	~7–15 days	<7 days
Ownership coverage	~60%	>90%
Exception aging	N/A	—
Tier C without mitigation plan	0	0

References: §28531380129

Generated by UK AI Operational Resilience · 117.9 AIC · ⌖ 10.5 AIC · ⊞ 5.2K · ◷

• expires on Jul 4, 2026, 8:23 AM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by UK AI Operational Resilience.

A newer discussion is available at Discussion #43013.