[security-observability] Daily Security Observability Report — 2026-08-28

[github-actions[bot]]

Executive Summary

This report combines two security signals over the last 7 days: firewall traffic analysis across 90 firewall-enabled workflow runs (61 unique workflows) and DIFC integrity-filtered event analysis for tool-call filtering. The vast majority of network traffic was benign — 5,125 requests allowed vs. 634 blocked (an ~11% block rate) — with most blocks concentrated in a handful of high-volume workflows performing package/dependency management (PureLock, CI Optimization Coach, Deep Report, Daily Go Test Parallelizer).

The top blocked domains are largely GitHub API and Go module proxy endpoints (api.github.com, proxy.golang.org, github.com), suggesting these workflows are hitting network egress restrictions during legitimate operations (API pagination, module fetching) rather than malicious activity. A long tail of ~80 domains was blocked exactly once each, almost entirely from a single run of the Deep Report workflow attempting broad package-manager and CA/OCSP connectivity — consistent with an overly restrictive allowlist for that workflow's toolchain rather than a security incident.

No DIFC integrity-filtered events were recorded in the last 7 days, indicating no tool calls were blocked by the Data Integrity and Flow Control system during this window — no runs matched the DIFC monitoring criteria, so this signal is clean with nothing to tune.

---

🔥 Firewall Analysis

Key Firewall Metrics

Metric	Value
Workflows analyzed (firewall-enabled)	90 (61 unique workflows)
Total network requests monitored	5,759
✅ Allowed requests	5,125
🚫 Blocked requests	634
Block rate	11.0%
Total unique blocked domains	86

📈 Firewall Request Trends

All analyzed runs fall within a single day (2026-08-28), so the chart shows hourly granularity. A pronounced spike in blocked requests occurs between 16:00–17:00 (407 and 149 blocked requests respectively), coinciding with the highest overall traffic volume of the day — most likely a batch of scheduled workflows (Go/dependency-heavy jobs) running concurrently. Outside this window, block volume stays low and steady (single digits to low tens per hour).

Top Blocked Domains

api.github.com (264 blocks) and proxy.golang.org (134 blocks) dominate, both from routine GitHub API and Go module operations rather than suspicious destinations. chatgpt.com/ab.chatgpt.com blocks (42 combined) appear across several AI-moderation-related workflows, likely blocked telemetry/tracking endpoints rather than intentional egress. No domains in the top 15 suggest malicious intent — this looks like firewall policy friction on legitimate developer tooling.

Most Frequently Blocked Domains

Domain	Times Blocked	Workflows	Category
api.github.com:443	264	Daily Go Test Parallelizer, Daily Max Ai Credits Test, Issue Monster, Ponytail Reviewer, PureLock	GitHub API
proxy.golang.org:443	134	CI Optimization Coach, Daily Formal Spec Verifier	Go module proxy
github.com:443	77	Contribution Check, Daily Go Test Parallelizer, Daily Malicious Code Scan Agent, Daily Max Ai Credits Test, Issue Monster, PR Sous Chef, Ponytail Reviewer, PureLock, UK AI Operational Resilience	GitHub web
chatgpt.com:443	27	AI Moderator, Daily Max Ai Credits Test, Issue Monster, Ponytail Reviewer	AI/LLM telemetry
(unknown)	19	AI Moderator, Architecture Guardian, Bot Detection, CLI Consistency Checker, Constraint Solving — Problem of the Day, Copilot PR Conversation NLP Analysis, Daily Copilot PR Merged Report, Daily Go Test Parallelizer, Daily Malicious Code Scan Agent, Daily Max Ai Credits Test, Daily Spending Forecast, Dependabot Dependency Checker, GitHub API Consumption Report Agent, GitHub MCP Structural Analysis, Issue Triage Agent, PureLock, Workflow Normalizer	Unresolved/DNS
ab.chatgpt.com:443	15	AI Moderator, Issue Monster, Ponytail Reviewer	AI/LLM telemetry
api.anthropic.com:443	13	Claude Code User Documentation Review, Copilot Agent Prompt Clustering Analysis, Deep Report, GitHub API Consumption Report Agent, GitHub MCP Structural Analysis, [aw] Failure Investigator (6h)	AI/LLM API
raw.githubusercontent.com:443	7	Daily Go Test Parallelizer, PureLock	GitHub raw content
*.grafana.net	1	Deep Report	Observability
*.pythonhosted.org	1	Deep Report	Python package hosting
*.sentry.io	1	Deep Report	Observability
anaconda.org	1	Deep Report	Package manager
api.anthropic.com	1	Deep Report	AI/LLM API
api.npms.io	1	Deep Report	npm search API
api.snapcraft.io	1	Deep Report	Snap package manager
archive.ubuntu.com	1	Deep Report	Ubuntu APT mirror
awmg-cli-proxy	1	Deep Report	Internal proxy
awmg-mcpg	1	Deep Report	Internal MCP gateway
binstar.org	1	Deep Report	Package manager (legacy Anaconda)

View Detailed Request Patterns by Workflow

Workflow	Allowed	Blocked
PureLock	274	247
CI Optimization Coach	102	133
Deep Report	28	84
Daily Go Test Parallelizer	121	69
Ponytail Reviewer	232	35
Daily Max Ai Credits Test	28	17
Issue Monster	30	10
AI Moderator	61	9
PR Sous Chef	148	3
UK AI Operational Resilience	64	3
GitHub API Consumption Report Agent	124	2
Contribution Check	90	2
Claude Code User Documentation Review	201	2
Daily Malicious Code Scan Agent	40	2
GitHub MCP Structural Analysis	92	2
Architecture Guardian	18	1
Copilot PR Conversation NLP Analysis	96	1
CLI Consistency Checker	88	1
Constraint Solving — Problem of the Day	28	1
(remaining ~43 workflows had 0–1 blocked requests each)	—	—

View Complete Blocked Domains List

api.github.com:443 (264), proxy.golang.org:443 (134), github.com:443 (77), chatgpt.com:443 (27), (unknown) (19), ab.chatgpt.com:443 (15), api.anthropic.com:443 (13), raw.githubusercontent.com:443 (7), and 78 additional domains blocked exactly once each (predominantly package-manager, OCSP/CRL certificate-validation, and Python/Node/Go tooling endpoints triggered by a single Deep Report run): *.grafana.net, *.pythonhosted.org, *.sentry.io, anaconda.org, api.anthropic.com, api.npms.io, api.snapcraft.io, archive.ubuntu.com, awmg-cli-proxy, awmg-mcpg, azure.archive.ubuntu.com, binstar.org, bootstrap.pypa.io, bun.sh, cdn.jsdelivr.net, conda.anaconda.org, conda.binstar.org, crl.geotrust.com, crl.globalsign.com, crl.identrust.com, crl.sectigo.com, crl.thawte.com, crl.usertrust.com, crl.verisign.com, crl3.digicert.com, crl4.digicert.com, crls.ssl.com, deb.nodesource.com, deno.land, esm.sh, files.pythonhosted.org, get.pnpm.io, googleapis.deno.dev, googlechromelabs.github.io, json-schema.org, json.schemastore.org, jsr.io, keyserver.ubuntu.com, nodejs.org, npm.pkg.github.com, npmjs.com, npmjs.org, ocsp.digicert.com, ocsp.geotrust.com, ocsp.globalsign.com, ocsp.identrust.com, ocsp.sectigo.com, ocsp.ssl.com, ocsp.thawte.com, ocsp.usertrust.com, ocsp.verisign.com, packagecloud.io, packages.cloud.google.com, packages.microsoft.com, pip.pypa.io, ppa.launchpad.net, pypi.org, pypi.python.org, registry.bower.io, registry.npmjs.com, registry.npmjs.org, registry.yarnpkg.com, repo.anaconda.com, repo.continuum.io, repo.yarnpkg.com, s.symcb.com, s.symcd.com, security.ubuntu.com, skimdb.npmjs.com, storage.googleapis.com, telemetry.vercel.com, ts-crl.ws.symantec.com, ts-ocsp.ws.symantec.com, www.googleapis.com, www.npmjs.com, www.npmjs.org, yarnpkg.com.

🔒 Firewall Security Recommendations

1. Investigate Deep Report egress needs — this single run generated ~80 unique single-hit blocked domains spanning npm, pip, conda, apt, and CA/OCSP validation endpoints. If this workflow legitimately needs broad package-manager access, consider an explicit allowlist bundle for its toolchain rather than relying on default deny; if not, confirm the blocks are intentional and not masking a misconfigured dependency step.
2. Review api.github.com / github.com block volume in PureLock, Ponytail Reviewer, and Daily Go Test Parallelizer — high block counts (247, 35, 69 total blocked requests respectively) against core GitHub endpoints may indicate rate-limiting-driven retries or overly narrow API scoping rather than a security concern; verify these workflows aren't failing silently due to blocked calls.
3. Resolve (unknown) domain classification (19 blocks across 17 workflows) — these are likely DNS resolution failures or non-HTTP egress; add logging/labeling improvements to the firewall proxy to identify what's actually being requested.
4. No policy rule-hit data was available in any audited run this period — policy_analysis was present in some audits but contained no rule_hits, so rule-level attribution isn't currently possible. Consider verifying the firewall proxy configuration to ensure rule-hit logging is enabled for future audits.
5. chatgpt.com/ab.chatgpt.com blocks (42 total) across AI-moderation workflows are consistent with blocked ChatGPT web telemetry, not the intended Anthropic/OpenAI API traffic — no action needed unless these workflows are unexpectedly relying on chatgpt.com directly.

---

🔒 DIFC Integrity Analysis

No DIFC integrity-filtered events found in the last 7 days. The DIFC data source (/tmp/gh-aw/agent/integrity/filtered-logs.json) returned zero runs, and the warm-start cache snapshot (/tmp/gh-aw/cache-memory/security-observability/filtered-logs.snapshot.json) confirms no filtered events in the prior period either. No tool calls were blocked by the Data Integrity and Flow Control system during this window — this is a clean signal with nothing to chart or tune.

---

Generated by the Daily Security Observability workflow (consolidated from Daily Firewall Reporter + Daily DIFC Analyzer)
Analysis window: Last 7 days | Repository: github/gh-aw
Run: https://github.com/github/gh-aw/actions/runs/33205444385

Generated by 🔒 Daily Security Observability Report · copilot · auto · 147.4 AIC · ⌖ 14.6 AIC · ⊞ 15.4K · ◷

• expires on Aug 31, 2026, 12:26 PM UTC-08:00

[github-actions[bot]]

This discussion has been marked as outdated by Daily Security Observability Report.

A newer discussion is available at Discussion #56917.