Borrowing it
Nothing to install: this file belongs to tbhb/vale-ai-tells. Take a copy, put it at the same path in your own repository, and replace the rules that are about this project with yours.
curl -O https://raw.githubusercontent.com/tbhb/vale-ai-tells/main/.claude/skills/codex-analyze-session/SKILL.mdgit clone --depth 1 https://github.com/tbhb/vale-ai-tellsWrote 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/skills/tbhb/vale-ai-tells/codex-analyze-session)<a href="https://agentmods.dev/skills/tbhb/vale-ai-tells/codex-analyze-session"><img src="https://agentmods.dev/badge/skills/tbhb/vale-ai-tells/codex-analyze-session/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/skills/tbhb/vale-ai-tells/codex-analyze-session"><img src="https://agentmods.dev/badge/skills/tbhb/vale-ai-tells/codex-analyze-session.svg" alt="Reviewed on agentmods" width="80" height="20"></a>- NVIDIA SkillSpector warn
SkillSpector: 1 finding, up to medium
These are SkillSpector’s own severities. On a checked sample its high-severity flags on skills were ~96% false positives — a documented command, a public API, a “never do X” rule — so we show them as a caution to read, not a verdict. Why →
- medium Rogue Agent · line 4 Skill establishes unauthorized persistence across sessions via cron jobs, startup scripts, or state files. Session persistence allows an attacker to maintain access beyond the current interaction.Fix: Remove any persistence mechanisms (cron jobs, startup scripts, state files). Skills should not maintain state across sessions without explicit user consent.
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 | $0.00079 | $0.01199 |
| Opus 5 | $0.00039 | $0.00600 |
| Sonnet 5 | $0.00016 | $0.00240 |
| Haiku 4.5 | $0.00008 | $0.00120 |
Grade A, and why
codex-analyze-session 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 10d 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 — 87 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Analyze a session
Work the steps in order. This skill ends once the retrospective exists on disk and the operator has its path. Acting on what it recommends is separate work. Use the codex-commit skill for any change that comes out of it.
Which session
Use the absolute transcript path from the user's request. A bare session ID also works when it names exactly one file under ~/.codex/sessions/.
If the user gives neither, preflight explains how to find a session. Spawn a fresh delegated agent with fork_turns: "none". Run preflight again with the path it returns. Never load session files into model context merely to locate one because the locator can stream them.
Preflight
Run bash .agents/skills/codex-analyze-session/scripts/preflight.sh <path-or-id> and read its report.
Step 0: open the task list
Create these tasks with update_plan, then move each to in_progress and completed as you go. They're the checklist the rest of this document expands.
- Read the lines the preflight listed
- Test the session's own conclusions
- Verify every figure with three agents
- Recommend fixes across the five areas
- Write the retrospective
Stop before any of it where preflight couldn't resolve the transcript, or reports a missing precondition. Say what's wrong and hand back.
The preflight already resolved the transcript and summarized the session. It created the retro directory at <main worktree>/.codex/retros/YYYYMMDD_<session ID>/ and copied the evidence into it. Read that report before forming any hypothesis, because it writes everything it prints to artifacts/ as well, where a later reader can check it.
.agents/skills/codex-analyze-session/references/SESSION_CLI.md documents every subcommand, for when a report needs more detail than the preflight printed.
Step 1: read the lines the preflight listed
The preflight listed the failing lines and the line ranges of each repeated sequence. Open those, and nothing else:
.agents/skills/codex-analyze-session/scripts/session.py grep <transcript> 'pattern' --context 300
.agents/skills/codex-analyze-session/scripts/session.py show <transcript> 361,408,1938
What ships with it
10 files beside SKILL.md in the same directory: the scripts, references and assets a skill reads on demand. Not counted in the per-session cost; read them before you install if any of them is executable.
- references/FIX_SURFACES.md 3.4 KB
- references/LOCATE_SESSION.md 2.4 KB
- references/SESSION_CLI.md 3.3 KB
- references/VERIFY_FIGURES.md 3.2 KB
- references/WRITING_THE_RETRO.md 3.1 KB
- scripts/new-retro.sh 4.5 KB runs code
- scripts/preflight.sh 5.4 KB runs code
- scripts/session.py 20 KB runs code
- templates/RETRO.md 2.6 KB
- tokens.json 1.2 KB
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.
- 10d ago First seen · 87 lines · 79 tokens per session scan A 736b99a3084d
codex-analyze-session is a skill published in the GitHub repository tbhb/vale-ai-tells (91 stars, last pushed today), licensed MIT. It adds 79 tokens to every session and 1,199 once invoked, about $0.0004 per session on Opus 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 2026-08-30.
Other skills, from other repositories
anti-slop
Remove AI slop from any voice-bearing prose — original posts, threads, articles, long-form, emails, docs, READMEs, marketing copy, bios, scripts. Use when drafting text meant to sound like a specific person or brand, when rewriting text that reads generic, corporate, or AI-generated, or when asked to humanize…
kill-slop
Audit a file or draft for AI slop and off-voice lines against the author's bound voice, report findings with in-voice swaps, then apply approved fixes. Use when the user says "kill slop", "/kill-slop [file]", "find the slop in this", "audit this for slop", or wants a deck, article, page, or draft cleaned to sound like…
humanizer
Humanize and de-slop AI-sounding prose while preserving meaning, specific detail, and genuine human quirks. When a voice sample or style profile is available, rewrite in that writer's actual voice. Remove recurring AI tells such as inflated significance, stock vocabulary, uniform rhythm, excessive hedging, formulaic…
authenticity-check
Score how authentically text reads as a real human author's work, flag spans that read as AI-generated, AI-templated, or generically derivative, and run a separate read-only scan for suspicious Unicode provenance carriers. Return an authenticity band, a 0-100 score, provenance signals, and span-level reasons. Use when…
ai-slop-detector
Universal prose audit. Scores writing on TWO axes — AI-Slop (does this read like AI wrote it?) and Comprehension (can a fresh reader follow this?). Use PROACTIVELY as a mandatory final QA pass on ANY prose generated for humans to read — every email (internal or external), proposal, report, status update, blog post…
humanizer-czech
Odstraň znaky AI-generovaného psaní z českého textu. Použij při editaci nebo revizi textu, aby zněl přirozeněji a lidštěji. Detekuje a opravuje 27 vzorců včetně: nafouklého významu, propagačního jazyka, anglického slovosledu, kalků, nominalizace, monotónního rytmu, nadužívání spojek, typických českých AI klišé…