ClawBlog

Live — Glass Newsroom

The newsroom watching itself.

Every ticket dispatched, draft submitted, QC score recorded, and article published by ClawBlog’s agents — with timestamps, capability tiers, and costs attributed. Public by construction per the principles in the ClawBlog Meta Column.

Open the Command Center — the ecosystem, mappedBrowse the Gallery — the newsroom’s generated imagery

Running in Lean Mode with auto-publish enabled. Drafts that pass automated QC and source verification publish without waiting for a human editor — each one is logged below as “Auto-published” with its QC score and verification counts. A human still curates sources, prompts, and corrections.

Events (7d)
132
Cost (7d)
$2.58
Posts published (7d)
6
QC rejections (7d)
2

Events / day

newsroom_events recorded · last 14 days

Oct 11Oct 18Oct 2414
Newsroom events recorded per day over the last 14 days Peaked at 30 on Oct 23.
PeriodValue
Oct 1126
Oct 1215
Oct 1313
Oct 1414
Oct 1513
Oct 1617
Oct 1714
Oct 1830
Oct 1916
Oct 2013
Oct 2114
Oct 2215
Oct 2330
Oct 2414

Cost / day

USD attributed to agent calls · last 14 days

Oct 11Oct 18Oct 240.3246
Agent cost in USD per day over the last 14 days Peaked at 0.6204 on Oct 23.
PeriodValue
Oct 110.323
Oct 120.2933
Oct 130.3142
Oct 140.3598
Oct 150.3127
Oct 160.3166
Oct 170.2855
Oct 180.5798
Oct 190.0437
Oct 200.3139
Oct 210.3571
Oct 220.3398
Oct 230.6204
Oct 240.3246

Caught this week

The publication’s misses, blocks, and saves over the last seven days — not the routine pipeline. Verification blocks, off-pack citation rejections, rotted links, and reader-tip promotions. The transparency promise lives or dies here.

Source packs built

10

Curated pools of pre-vetted URLs handed to the writer.

Verification blocks

0

Pre-publish re-fetch failed on a cited URL.

Off-pack citations rejected

0

Writer tried to cite a URL outside the source pack.

Rotted links surfaced

1

Freshness sweep found a citation that no longer resolves.

Tips promoted

0

Reader-submitted URL ended up in a published article.

  1. Source packs built

    Source pack built — 12/31 items

  2. Source packs built

    Source pack built — 11/29 items

  3. Source packs built

    Source pack built — 13/32 items

  4. Source packs built

    Source pack built — 15/34 items

  5. Source packs built

    Source pack built — 12/32 items

  6. Source packs built

    Source pack built — 10/30 items

Recent activity

  1. Hero Imagekernel

    Hero image generated for "Kimi K3.1 Pro Hits Fable-Class Ten Weeks Early, and the Closed-Model Lead Shrinks to About a Month"

    $0.06
    Structured detail
    {
  "model": "google/gemini-nano-banana-2.1",
  "provider": "openrouter",
  "resolved_model": "google/gemini-nano-banana-2.1",
  "_redacted_keys": 1
}
  2. Hero Queuedkernel

    Hero image sync-gen attempt for "Kimi K3.1 Pro Hits Fable-Class Ten Weeks Early, and the Closed-Model Lead Shrinks to About a Month" (google/gemini-nano-banana-2.1)

    Structured detail
    {
  "model": "google/gemini-nano-banana-2.1",
  "via_queue": false
}
  3. Completedcron

    Cron tick — longform draft ingested

    Structured detail
    {
  "scout_chain": [
    {
      "stage": "scout-initial",
      "reason": "promoted supported runner-up (2 in-pack sources)",
      "outcome": "picked"
    }
  ],
  "_redacted_keys": 4
}
  4. Auto-Publishedcron

    Auto-published — QC signed off at 80 (full-auto: QC approval is the gate), 5/5 URLs verified

    Structured detail
    {
  "post_id": "316",
  "qc_score": 80,
  "urls_total": 5,
  "publish_gate": "full-auto-qc-signoff",
  "urls_verified": 5,
  "_redacted_keys": 1
}
  5. Verifiedsource-pack

    Verification passed — 5/5 URLs verified

    Structured detail
    {
  "total": 5,
  "verified": 5,
  "_redacted_keys": 4
}
  6. Claim Groundingkernel

    Claim grounding 83% across 7 bound claim(s)

    Structured detail
    {
  "_redacted_keys": 3
}
  7. QC Scoredqc-editor· a382645d-90cf-4dbe-b1fa-af210b32418d

    QC score 80 — approved

    tier2_reasoningdeepseek/deepseek-v4.1-flash$0.01

    QC reasoning · auto

    Depth
    8
    Voice
    7
    Sources
    8
    Accuracy
    8
    Structure
    8
    Readability
    8
    Seo Hygiene
    7
    Framework Fit
    6
  8. Draftscout+writer· tide· a382645d-90cf-4dbe-b1fa-af210b32418d

    Draft submitted: Kimi K3.1 Pro Hits Fable-Class Ten Weeks Early, and the Closed-Model Lead Shrinks to About a Month

    tier1_budget_generationanthropic/claude-opus-5.5$0.23
    Structured detail
    {
  "pillar": "news",
  "word_count": 2093,
  "_redacted_keys": 1
}
  9. Dispatchedcron

    Cron tick — transient-failure retry (attempt 2/3)

    tier1_budget_generation
    Structured detail
    {
  "retry": "transient-retry",
  "_redacted_keys": 2
}
  10. Costcron

    Scout scout-initial — $0.0146

    tier1_budget_generation$0.01
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-initial",
  "_redacted_keys": 1
}
  11. Scout Dispatchedcron

    Scout pass — initial angle search

    tier1_budget_generation
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-initial"
}
  12. Pack Builtsource-pack

    Source pack built — 12/31 items

    Structured detail
    {
  "stats": {
    "fetched": 31,
    "accepted": 12,
    "rejected": 19,
    "candidates": 31,
    "rejectedReasons": {
      "http-403": 1,
      "unknown-tier-off-topic": 18
    }
  },
  "pack_id": 337,
  "top_reasons": [
    {
      "count": 18,
      "reason": "unknown-tier-off-topic"
    },
    {
      "count": 1,
      "reason": "http-403"
    }
  ]
}
  13. Dispatchedcron

    Cron tick — 31 candidate item(s) (3 URL-filtered + 1 semantic-deduped as recently covered)

    tier1_budget_generation
    Structured detail
    {
  "count": 31,
  "filtered_semantic": 1,
  "semantic_duplicates": [
    {
      "title": "Research write-up + runnable lab on MCP discovery-time prompt injection",
      "reason": "same MCP security vulnerability (prompt injection via discovery-time instructions field)",
      "matched": "Pydantic-AI's MCP Telemetry Advisory Is Rated Medium. The Boundary It Exposes Is Not."
    }
  ],
  "filtered_recently_covered": 3
}
  14. Costcron

    Semantic dedup — $0.0050

    tier1_budget_generation$0.0050
    Structured detail
    {
  "model": "semantic-dedup-haiku",
  "_redacted_keys": 1
}
  15. Hero Imagekernel

    Hero image generated for "Show It Once: Generalist's Promptable Robots Bet Against the Retraining Treadmill"

    $0.06
    Structured detail
    {
  "model": "google/gemini-nano-banana-2.1",
  "provider": "openrouter",
  "resolved_model": "google/gemini-nano-banana-2.1",
  "_redacted_keys": 1
}
  16. Hero Queuedkernel

    Hero image sync-gen attempt for "Show It Once: Generalist's Promptable Robots Bet Against the Retraining Treadmill" (google/gemini-nano-banana-2.1)

    Structured detail
    {
  "model": "google/gemini-nano-banana-2.1",
  "via_queue": false
}
  17. Completedcron

    Cron tick — longform draft ingested

    Structured detail
    {
  "scout_chain": [
    {
      "stage": "scout-initial",
      "reason": "Pack is trade-rag-only and/or off-beat. Items 1-2 and 11 are routine framework release notes (langgraph CLI/SDK version bumps, langfuse feature releases) — developer-plumbing churn with no consumer-agent user-facing impact, banned per hard-ban rules. Item 3 (Generalist robotics) is on-beat but niche robotics subject, not core consumer-agent stack. Item 4 is a meta-update about Latent Space's platform migration, not substantive agent news. Items 5, 9, 10 are Stratechery / general tech, entirely off-beat. Item 6 is a music video. Item 7 is a throwaway quote. Item 8 (Moshi) is a trading-specific tool — vertical domain, narrow audience interest, fails the 'would Wired/Verge/NYT-Tech reader recognize this' test. No item clears the bar for consumer-agent notability, forward-looking substance, or",
      "outcome": "no-pick"
    },
    {
      "stage": "scout-force-pick",
      "outcome": "picked"
    }
  ],
  "_redacted_keys": 4
}
  18. Auto-Publishedcron

    Auto-published — QC signed off at 79 (full-auto: QC approval is the gate), 6/6 URLs verified

    Structured detail
    {
  "post_id": "315",
  "qc_score": 79,
  "urls_total": 6,
  "publish_gate": "full-auto-qc-signoff",
  "urls_verified": 6,
  "_redacted_keys": 1
}
  19. Verifiedsource-pack

    Verification passed — 6/6 URLs verified

    Structured detail
    {
  "total": 6,
  "verified": 6,
  "_redacted_keys": 4
}
  20. Claim Groundingkernel

    Claim grounding 86% across 6 bound claim(s)

    Structured detail
    {
  "_redacted_keys": 3
}
  21. QC Scoredqc-editor· 7aa5eb38-bfbc-4623-b4d0-958d04130df1

    QC score 79 — approved

    tier2_reasoningdeepseek/deepseek-v4.1-flash$0.02

    QC reasoning · auto

    Depth
    6
    Voice
    7
    Sources
    7
    Accuracy
    8
    Structure
    9
    Readability
    8
    Seo Hygiene
    6
    Framework Fit
    7
  22. Draftscout+writer· reef· 7aa5eb38-bfbc-4623-b4d0-958d04130df1

    Draft submitted: Show It Once: Generalist's Promptable Robots Bet Against the Retraining Treadmill

    tier1_budget_generationanthropic/claude-opus-5.5$0.22
    Structured detail
    {
  "pillar": "deep-dives",
  "word_count": 2320,
  "_redacted_keys": 1
}
  23. Costcron

    Scout scout-force-pick — $0.0144

    tier1_budget_generation$0.01
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-force-pick",
  "_redacted_keys": 1
}
  24. Scout Dispatchedcron

    Scout pass — force-pick retry (initial pass returned no-pick)

    tier1_budget_generation
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-force-pick"
}
  25. Costcron

    Scout scout-initial — $0.0109

    tier1_budget_generation$0.01
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-initial",
  "_redacted_keys": 1
}
  26. Scout Dispatchedcron

    Scout pass — initial angle search

    tier1_budget_generation
    Structured detail
    {
  "model": "scout-pass-haiku",
  "stage": "scout-initial"
}
  27. Pack Builtsource-pack

    Source pack built — 11/29 items

    Structured detail
    {
  "stats": {
    "fetched": 29,
    "accepted": 11,
    "rejected": 18,
    "candidates": 29,
    "rejectedReasons": {
      "http-403": 1,
      "unknown-tier-off-topic": 17
    }
  },
  "pack_id": 336,
  "top_reasons": [
    {
      "count": 17,
      "reason": "unknown-tier-off-topic"
    },
    {
      "count": 1,
      "reason": "http-403"
    }
  ]
}
  28. Dispatchedcron

    Cron tick — 29 candidate item(s) (3 URL-filtered + 3 semantic-deduped as recently covered)

    tier1_budget_generation
    Structured detail
    {
  "count": 29,
  "filtered_semantic": 3,
  "semantic_duplicates": [
    {
      "title": "Research write-up + runnable lab on MCP discovery-time prompt injection",
      "reason": "same MCP security vulnerability; discovery-time prompt injection is a follow-up development to the telemetry advisory",
      "matched": "Pydantic-AI's MCP Telemetry Advisory Is Rated Medium. The Boundary It Exposes Is Not."
    },
    {
      "title": "Introducing Machine Gods",
      "reason": "PHP2Shell is the same vulnerability incident referenced in the existing PHP bug coverage",
      "matched": "Two Teams, One Model, Same PHP Bug: What OpenAI's Patch the Planet Reveals About AI Vulnerability Hunting"
    },
    {
      "title": "Why AI Compute Comes in Rectangles — Erik Bernhardsson, Modal CEO",
      "reason": "same company and funding/strategic positioning already covered",
      "matched": "Modal's $16B Series D Says the Agent Money Is in Sandboxes, Not Models"
    }
  ],
  "filtered_recently_covered": 3
}
  29. Costcron

    Semantic dedup — $0.0057

    tier1_budget_generation$0.0057
    Structured detail
    {
  "model": "semantic-dedup-haiku",
  "_redacted_keys": 1
}
  30. Completednewsletter-agent

    Weekly digest sent — 5 articles

    Structured detail
    {
  "subject": "ClawBlog Newsletter — Oct 16 – Oct 23, 2026",
  "_redacted_keys": 1
}

Data layer: newsroom_events. Read-only public view. The redaction layer (Phase 4C) hides detail payloads flagged by the emitting agent.