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 skills add Marcel-Bich/marcel-bich-claude-marketplace --skill requirements-verbatimgit clone --depth 1 https://github.com/Marcel-Bich/marcel-bich-claude-marketplaceWrote 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/marcel-bich/marcel-bich-claude-marketplace/requirements-verbatim)<a href="https://agentmods.dev/skills/marcel-bich/marcel-bich-claude-marketplace/requirements-verbatim"><img src="https://agentmods.dev/badge/skills/marcel-bich/marcel-bich-claude-marketplace/requirements-verbatim/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/marcel-bich/marcel-bich-claude-marketplace/requirements-verbatim"><img src="https://agentmods.dev/badge/skills/marcel-bich/marcel-bich-claude-marketplace/requirements-verbatim.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 34 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 | $0.00101 | $0.01310 |
| Opus 5 | $0.00051 | $0.00655 |
| Sonnet 5 | $0.00020 | $0.00262 |
| Haiku 4.5 | $0.00010 | $0.00131 |
Grade A, and why
requirements-verbatim 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 12d 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 — 116 lines — stays where its author put it; the contents beside it link to each section on GitHub.
requirements-verbatim
Record requirements, decisions, approvals, and constraints exactly as the user stated them, into a durable on-disk log. The log is the ground truth for what was asked. It outranks memory, paraphrase, and model inference (see the authority order below). Its single job: nothing the user approved can be lost, altered, or quietly reinterpreted between now and when the work is verified.
When this fires
- The user states a requirement, gives a GO, sets a constraint, or makes a decision.
- The user approves or rejects a proposal (log what was approved, verbatim).
- You (or a subagent) are handed a requirement and are about to act on it: log it first.
- Before any compaction (see the credo
compact-plusskill): the verbatim log must already hold everything approved, because a compact must not be able to lose or alter it.
This fires inside subagents too. A subagent that receives a requirement from the main agent logs the verbatim source itself rather than trusting a paraphrase.
Where it lives
Append-only log files under .credo/process/requirements/ in the current project
(the credo per-project namespace created by scripts/credo-init.sh). One dated file,
for example .credo/process/requirements/2026-07-04.md, appended to across the day.
- Append only. Never edit or delete an existing entry. A correction is a NEW entry that references the earlier one; the original stays.
- Write atomically (temp file plus
mv -f) per the credo persistence convention, so a crash mid-write cannot corrupt the log. .credo/**is intentionally git-excluded. Durability comes from disk plus whatever external backup the environment provides (for example an automatic file backup), not from commits. Do not add or force-add.credo/to git.
Verbatim fidelity (anti-censorship - absolute)
- Quote the user WORD FOR WORD. Preserve wording, ordering, emphasis, and any explicit numbers or limits.
- NEVER trim, soften, summarize away, reinterpret, sanitize, or censor what the user said. This holds regardless of the content, including sensitive or uncomfortable material. If in doubt, quote more, not less.
- NEVER invent a constraint the user did not state. Do not present your own interpretation as the user's requirement.
- Keep user-verbatim STRICTLY separate from anything you add. Every entry has two
clearly labeled zones:
User (verbatim)- the exact words, quoted. Nothing of yours here.Assistant (proposal / finding)- optional. Your interpretation, proposal, open question, or note. Clearly marked as yours, never mixed into the verbatim zone.
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.
- 12d ago First seen · 116 lines · 101 tokens per session scan A 6ae698c45288
requirements-verbatim is a skill published in the GitHub repository Marcel-Bich/marcel-bich-claude-marketplace (13 stars, last pushed yesterday), licensed MIT. It adds 101 tokens to every session and 1,310 once invoked, about $0.0005 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
customer-success
Support workflows, ticketing systems (Zendesk, Intercom), knowledge base design, chatbot design, and metrics (CSAT, NPS). Use when building support infrastructure, designing help centers, or optimizing customer experience.
event-planner
Event planning with timelines, budgets, vendor coordination, logistics checklists, and post-event evaluation. Use when organizing conferences, workshops, galas, or corporate events.
hr-talent
HR and talent management expertise for talent acquisition, performance management, compensation strategy, organizational design, culture building, succession planning, and D&I programs. Use when hiring, managing performance, designing organizations, or building culture.
innovation
Innovation management expertise for innovation frameworks (Design Thinking, Stage-Gate), ideation processes, innovation portfolio management, venture capital, open innovation, and IP strategy. Use when driving innovation, managing R&D portfolios, or building innovation programs.
agent-teams
Team composition knowledge for Claude Code Agent Teams - when to suggest teams, optimal sizing, spawn prompt patterns.
execute
Execute phase — wave-based crew dispatch with critic gates and evidence-based verification.