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 commands/dimitrigeelen/agentic-engineering-framework/writegit clone --depth 1 https://github.com/DimitriGeelen/agentic-engineering-frameworkWrote 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/commands/dimitrigeelen/agentic-engineering-framework/write)<a href="https://agentmods.dev/commands/dimitrigeelen/agentic-engineering-framework/write"><img src="https://agentmods.dev/badge/commands/dimitrigeelen/agentic-engineering-framework/write.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 | $0.00569 |
| Opus 5 | $0.00000 | $0.00284 |
| Sonnet 5 | $0.00000 | $0.00114 |
| Haiku 4.5 | $0.00000 | $0.00057 |
Grade A, and why
write 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 today.
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 — 63 lines — stays where its author put it; the contents beside it link to each section on GitHub.
/write - Voice-Calibrated Content Production
Produce content drafts calibrated to the project's writing voice. Use for articles, posts, documentation, and any public-facing prose.
Prerequisites
- Verify an active task exists (
cat .context/working/focus.yaml) - Read the style guide:
docs/style-guide.md - Read the active task file for deliverable specs and ACs
Workflow
Step 1: Load Voice
Read docs/style-guide.md. Internalize the voice characteristics and anti-patterns
as creative constraints for this session.
Step 2: Understand the Deliverable
From the task file, extract:
- Target platform (dev.to, LinkedIn, Reddit, docs)
- Word count target (from ACs or description)
- Key messages or claims to include
- Output file path (from ACs)
If the task does not specify a platform, ask before drafting.
Step 3: Draft
Write the content to the specified output path, applying:
- All 10 voice characteristics from the style guide
- The 5 translation rules (principle-first opening, cross-domain parallels, pulled quotes, medium paragraphs, experience-grounded authority)
- Platform-specific adjustments from the style guide table
- The task's acceptance criteria as structural requirements
Step 4: Self-Check
Before presenting the draft, verify against these mechanical tests:
- First paragraph test: Does it work if the reader has never used an AI coding agent?
- Pulled quote test: Are there 1-2 standalone bold sentences in domain-neutral language?
- "You" test: Is second-person scenario setup replaced with third-person cross-domain parallels?
- Paragraph test: Are there any one-sentence paragraphs? (Delete or weave in.)
- Authority test: Does any paragraph open with a statistic? (Move it later.)
- Anti-pattern scan: No emojis, no exclamation marks, no "we," no rhetorical questions in isolation, no "Let me show you," no headers as questions.
Report to the user:
- File path and word count
- Which platform adjustments were applied
- Any ACs that could not be satisfied in draft
- Self-check results (pass/fail per test)
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.
- today First seen · 63 lines · 0 tokens per session scan A 5f68b786e762
write is a command published in the GitHub repository DimitriGeelen/agentic-engineering-framework (13 stars, last pushed today), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 569 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 2026-09-04.
Other commands, from other repositories
service-check
Request each agent to run a self-diagnostic to validate their runtime setup.
compatibility-audit
Audit all agents for Trinity compatibility by requesting self-diagnostics from each agent.
request-research
Ask the researcher to investigate a specific topic.
health
Perform a health check on all agents and identify any issues.
schedules
Quick overview of all schedules across the platform.
status
Report on recent research activity and findings count.