Getting it into your agent
One page per mod, every tool's command on it. A separate URL per tool would split the same page into five that compete with each other.
npx agentmods add agents/gaelic-ghost/swiftasb/codex-direct-read-plangit clone --depth 1 https://github.com/gaelic-ghost/SwiftASBWrote 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/gaelic-ghost/swiftasb/codex-direct-read-plan)<a href="https://agentmods.dev/agents/gaelic-ghost/swiftasb/codex-direct-read-plan"><img src="https://agentmods.dev/badge/agents/gaelic-ghost/swiftasb/codex-direct-read-plan.svg" alt="Measured on agentmods" 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 | $0.00000 | $0.00968 |
| Opus 5 | $0.00000 | $0.00484 |
| Sonnet 5 | $0.00000 | $0.00194 |
| Haiku 4.5 | $0.00000 | $0.00097 |
Grade A, and why
codex-direct-read-plan 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 3d 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 — 114 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AgentSB Direct Read Prototype Plan
AgentSB may use direct local Codex reads as an experimental maintainer lane for
faster reports and for prototyping the future SwiftASB direct-thread-storage
feature. This lane is private, read-only by default, and version-gated. The
destination feature is tracked in
../maintainers/codex-direct-thread-storage-plan.md.
Goal
Use local Codex storage to help design and validate a faster SwiftASB thread inventory path without clogging the CLI/app-server JSONL pipe. The first AgentSB target is thread inventory: read SQLite metadata first, then lazily read referenced JSONL only when deeper turn evidence is needed.
Supported Storage Contract
The first experimental contract targets the local Codex storage shape observed on 2026-06-04:
~/.codex/state_5.sqlite- table:
threads - rollout JSONL paths under
~/.codex/sessions/and~/.codex/archived_sessions/ - active and archived JSONL using the same broad envelope families:
session_meta,turn_context,event_msg, andresponse_item
AgentSB must treat this as a prototype compatibility window, not a forever-stable schema. If the table, required columns, or path conventions drift, AgentSB should report the mismatch and stop before reading deeper content.
Direct Read Principles
- Open SQLite databases read-only.
- Never mutate Codex state, archive flags, JSONL files, logs, goals, or app support files.
- Prefer metadata summaries over prompt/tool-content ingestion.
- Redact private text by default, including titles, first user messages, and previews.
- Require an explicit flag before including private title, prompt, or preview text.
- Read JSONL lazily and only for selected rows.
- Keep direct-read outputs labeled as private local Codex storage observations.
- Keep app-server surfaces authoritative for product behavior, archive actions, live state, and compatibility claims while SwiftASB's direct-read design is still experimental.
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.
- 3d ago First seen · 114 lines · 0 tokens per session scan A 810a04de58fb
codex-direct-read-plan is an agent published in the GitHub repository gaelic-ghost/SwiftASB (4 stars, last pushed 1mo ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 968 tokens. 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
spark-pod-autosizing
Optimize Apache Spark workload configurations with AI-powered resource recommendations based on historical usage patterns. Get rightsized CPU, memory, and storage recommendations for Spark drivers and executors.
omp
Install Remarc's public OMP plugins, verify the MCP connection, and pair a session for instant comment delivery.
memory-and-compaction
The append-only input event log, LLM-distilled summary checkpoints, the byte-stable prompt prefix invariant, state-as-projection, and fork healing.
mcp-tools-reference
Reference for the seven MCP tools Remarc exposes to agents, with status strings, filters, and deliberate limits.
cursor
Enable Remarc's app-managed Cursor integration; the app writes the MCP config and skill for you and uninstalls them on toggle-off.
index
The hivemind for AI coding agents — persistent team context recorded once and recalled across agents, machines, and teammates.