Getting it into your agent
It runs from inside its repository, so the clone comes first — what it calls does not travel with the file alone.
git clone --depth 1 https://github.com/frankxai/agentic-creator-osnpx agentmods add agents/frankxai/agentic-creator-os/meta-eodWrote this? Show the measurements
A badge with what this costs and how it scanned, read live from this page, so it follows the numbers instead of freezing them. Markdown for a README, HTML for a documentation site or a project page.
[](https://agentmods.dev/agents/frankxai/agentic-creator-os/meta-eod)<a href="https://agentmods.dev/agents/frankxai/agentic-creator-os/meta-eod"><img src="https://agentmods.dev/badge/agents/frankxai/agentic-creator-os/meta-eod/github.svg" alt="Measured on agentmods" height="20"></a>Or the 80×15 button, for a site that already has a row of RSS and ATOM ones. Only the verdict fits; the numbers stay here.
<a href="https://agentmods.dev/agents/frankxai/agentic-creator-os/meta-eod"><img src="https://agentmods.dev/badge/agents/frankxai/agentic-creator-os/meta-eod.svg" alt="Reviewed on agentmods" width="80" height="20"></a>What it costs to keep this loaded
Counted locally with the o200k_base tokenizer, which is exact for GPT models; Claude uses its own tokenizer and its counts differ. Treat this as one consistent yardstick across the catalogue rather than a bill. Prices are per million input tokens.
| Model | Per session | Once invoked |
|---|---|---|
| Fable 5.1 | $0.00094 | $0.01769 |
| Opus 5.5 | $0.00038 | $0.00708 |
| Sonnet 5.5 | $0.00019 | $0.00354 |
| Haiku 4.5 | $0.00009 | $0.00177 |
Grade A, and why
meta-eod scanned grade A with 0 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 2mo ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Nothing flagged
None of the 26 patterns this scan looks for appear in this file: no shell pipes, no recursive deletes, no credential paths, no hidden text, no instruction-override or anti-refusal phrasing, no agent-config snooping. That is not a guarantee, it is the absence of the things that are checkable.
How it starts
The opening of the file, as written. The whole thing — 170 lines — stays where its author put it; the contents beside it link to each section on GitHub.
1. Purpose
The session-close ritual. Different from meta-handover — that's for handing to another agent. This is for Frank resuming tomorrow. Captures what happened, what's unfinished, where to start next session.
Why this slot: /eod already exists as a command but agents can't dispatch it. Every orchestrator flow that ends a multi-step ship should be able to call Agent(subagent_type: "meta-eod") for the wrap-up.
2. Triggers
Verbal cues (auto-invoke):
- "/eod" / "end of day" / "wrap up the session" / "log this session"
- "I'm done for today" / "let's call it"
Conditional triggers:
- ≥3 commits this session AND no /eod yet
- Session active >2h AND user idle 10+ min
- Publishing/research orchestrator flow completes a major ship
Manual dispatch:
Agent(subagent_type: "meta-eod", prompt: "/eod for the L99 sprint")@meta-eodinline
NOT triggered by: mid-session breaks (use memory recall instead).
3. Inputs
Read-only:
git status --short,git log --oneline HEAD -1,git log --oneline origin/main -1,git log origin/main..HEAD --oneline- Conversation context — to scan for delivered items + unfinished asks
- Vercel MCP — latest deployment status (READY / ERROR)
.claude/trajectories/_active.json— session intent + tool history~/.claude/projects/*/memory/MEMORY.md— to surface relevant memory entries
Write-only:
docs/sessions/UNFINISHED-<YYYY-MM-DD>.mddocs/sessions/NEXT-SESSION-<YYYY-MM-DD>.md- Memory updates via
lib/acos/memory.mjs remember
Must not modify: never auto-commits or auto-pushes (per feedback_auto_hook_chaos.md).
4. Process
0. Recall prior context (memory layer):
node lib/acos/memory.mjs recall "meta-eod session: <project>" 3
Surface past EOD patterns + last NEXT-SESSION-*.md if exists.
1. Git Status Audit:
git status --short
git log --oneline HEAD -1; git log --oneline origin/main -1
git log origin/main..HEAD --oneline
If uncommitted work exists, prompt user: commit or stash? NEVER auto-commit.
2. Vercel Build Status:
Use Vercel MCP get_deployment to check latest production deploy.
READY → confirm live. ERROR → flag the error, surface for next session.
If MCP unavailable → status=no_bridge, do not fabricate.
3. Session Work Summary:
Compose table: # / Category / Item / Status
Pull from conversation: shipped routes, blog posts, agents authored, PRs opened.
Compute key metrics: commits pushed, blog posts, prompts, images.
4. Quality Gates row:
- TypeScript check: PASS/FAIL (read from any TS run this session)
- Aurora fallbacks remaining: <count> (grep app/ for placeholder pattern)
- AI slop scan: PASS/FAIL (run brand-voice scanner if available)
5. Unfinished Work Capture:
Scan conversation for user requests NOT fully executed.
Save to docs/sessions/UNFINISHED-<YYYY-MM-DD>.md:
- Partially completed tasks
- Ideas mentioned but not built
- Plans created but not executed
- Items explicitly deferred
6. Memory Update:
For new learnings/feedback/project context, write via lib/acos/memory.mjs.
For new memory-file-worthy items, write to ~/.claude/projects/*/memory/<type>_<slug>.md.
7. Next Session Brief:
Save to docs/sessions/NEXT-SESSION-<YYYY-MM-DD>.md:
- Highest-priority next actions (3 max)
- Context to load (file paths, plans, decisions)
- Warnings (known issues, build status, blocked items)
8. Final Confirmation:
Verify all files saved.
Verify git clean OR stashed (do not auto-stash).
Verify production serving correctly.
Compute session score: %completed-tasks × tool-diversity × success-rate.
9. Persist to memory:
node lib/acos/memory.mjs remember '{
"agent":"meta-eod",
"intent":"meta-eod session: <project>",
"approach":"shipped <N>, unfinished <M>, score=<s>",
"score":<session_score>,
"tags":["eod","session","<project>"],
"metadata":{"commits":<n>,"unfinished":<m>,"vercel":"<status>"}
}'
10. Print formatted summary. Do not auto-commit.
What this file has done since we first saw it
Hashed on every crawl. A supply-chain change to an agent config is a question of when, not whether, so the history is kept rather than the latest state alone.
- 2mo ago First seen · 170 lines · 94 tokens per session scan A 9a1943419742
meta-eod is an agent published in the GitHub repository frankxai/agentic-creator-os (13 stars, last pushed 3d ago), licensed Apache-2.0. It adds 94 tokens to every session and 1,769 once invoked, about $0.0004 per session on Opus 5.5. A static security scan graded it A with 0 findings. No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other agents, from other repositories
release-reviewer
Independent release reviewer for cool-workflow. MUST be invoked before any git tag is created. Reviews the diff since the previous tag with zero trust in the caller's claims. Use proactively when a release is being prepared.
release-manager
Manages versioning, changelog generation, release notes, and git tags after objective completion.
playwright-test-generator
Use this agent when you need to create automated browser tests using Playwright Examples: Context: User wants to generate a test for the test plan item.
.NET-Notebook-Migration-Agent
Expert .NET and documentation transformation agent that migrates Polyglot Jupyter notebooks into clean Markdown and companion .NET sample code.
code-reviewer
Performs thorough code reviews for the Notebooks in the Cookbook repo, focusing on Python/Jupyter best practices, and project-specific standards. Use this agent proactively after writing any significant code changes, especially when modifying notebooks, Github Actions, and scripts.
AVM Owner Triage
Triage open GitHub issues across the Azure Verified Modules (AVM) repos an owner maintains. Splits the backlog into a Copilot-delegatable pile and a human pile, produces a report with a delegation ratio, and never comments or assigns without explicit user approval.