[uk ai resilience] [uk-ai-resilience] Open-code risk report - 2026-10-14

[github-actions[bot]]

Executive summary

Lookback 7 days (since 2026-10-07): 389 commits (229 security-signal), 21 open security issues, 249 open code-scanning alerts, 0 secret-scanning alerts. Alert mix: 2 critical, 236 high, 6 medium, 5 unrated. 219 high alerts are one pattern (non-deterministic npm install in workflows); 10 more are uv/pip. These are already tracked by #68043. No secret exposure. Note: the sub-agents were not dispatched; this assessment was done directly from pre-computed data, so confidence is moderate.

Asset graph summary

Recent-change scoped asset graph

Segment	Signal	Evidence
pkg/workflow (compiler, maintenance workflow YAML)	2 critical go/unsafe-quoting, go/weak-sensitive-data-hashing, go/bad-redirect-check	maintenance_workflow_yaml_jobs.go:634,668; expression_extraction.go:423
pkg/cli	graphql sprintf injection, integer conversion, bad-redirect-check	project_command.go:269,272; outcome_eval_review.go:143
.github/workflows/*.lock.yml	229 unpinned/non-deterministic installs	tracked in #68043
.github/scripts (JS)	insecure temp file, file-system race, URL substring sanitization	lint_monster_issues.cjs:66, test_secure_write.cjs:119, test_aw_issue_clustering_publish.cjs:277
Container images	CVE burn-down issues #68340-#68347, #52657	container-image-scan
Workflow static analysis	RGS-018 payload execution #68606-#68608	static-analysis

Ownership signals: top committers are Copilot (173), Peli de Halleux (163), github-actions[bot] (48).

Tier classification

Tier table

Area	Tier
Compiler / maintenance workflow generation (pkg/workflow)	B
CLI (pkg/cli)	B
Workflow lock files (npm/pip/uv installs)	B (tracked #68043)
JS scripts/tests (.github/scripts)	B
Container images	B
Docs/site scripts	A

No Tier C/D areas identified.

Control verification gaps

• Ownership: no CODEOWNERS-level evidence reviewed in this run; bot-heavy authorship lowers ownership confidence.
• SDLC: 249 open code-scanning alerts with a large backlog of unrated/high alerts; triage velocity is the gap.
• Dependency: 229 non-deterministic installs; container CVE burn-down issues open.
• Secrets: none open (0 alerts) – control effective.
• Runtime observability: not assessed.
• Recovery: patch path is regenerate-and-recompile lock files; rollback readiness not evidenced.

Risk scoring

Scores (1 low risk – 5 high risk)

Area	Exposure	Patchability	Detectability	Fragility	Ownership	Priority
unsafe-quoting in maintenance jobs	3	2	2	3	2	high (likely false positive, needs triage)
lock-file installs	4	2	1	3	2	high (#68043)
graphql sprintf in project_command.go	3	2	2	2	2	medium
JS script alerts	2	1	1	1	2	medium

Remediation queue

Action	SLA
Triage the 2 critical go/unsafe-quoting alerts (#1030, #1031); fix or dismiss with justification	high
Continue #68043 deterministic installs	high
Review graphql sprintf in project_command.go	medium
Triage remaining high JS/Go alerts	medium

Human-review trigger: any critical alert dismissal.

Exception register

None.

Operational metrics baseline

• MTTR proxy: not computed (no closure data in inputs).
• Ownership coverage: not measured.
• Unsupported dependency ratio: not measured; 229 non-deterministic installs flagged.
• Exception aging: n/a (0 exceptions).
• Exposure without recovery capability: none evidenced.

Generated by UK AI Operational Resilience · copilot · auto · 16.4 AIC · ⌖ 0.729 AIC · ⊞ 8K · ◷

• expires on Oct 17, 2026, 7:39 AM UTC-08:00