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/davidmatousek/tachi/readmegit clone --depth 1 https://github.com/davidmatousek/tachiWhat 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 | $0.00432 |
| Opus 5 | $0.00000 | $0.00216 |
| Sonnet 5 | $0.00000 | $0.00086 |
| Haiku 4.5 | $0.00000 | $0.00043 |
Grade A, and why
README 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.
What it actually says
Agents — Knowledge System Personas
Pattern: Domain-Specific Agent Personas
Agent personas are focused specialists that commands invoke to execute domain tasks. Each persona has a clear expertise boundary and knows which content to reference.
How to Design Agent Personas
-
Identify expertise boundaries — Each agent knows one aspect of your domain. A drafting agent knows voice and content; a review agent knows the scoring rubric; an export agent knows output formats.
-
One file per persona — Create persona files in this directory using kebab-case (e.g.,
resume-drafter.md,chapter-reviewer.md). -
Define what the agent knows — Each persona file specifies:
- What content the agent reads (VoiceProfile, specific MasterContent, Terms)
- What quality criteria the agent applies
- What anti-patterns the agent watches for
-
Map agents to commands — Document which commands invoke which agents. A command may use multiple agents; an agent may serve multiple commands.
Example Structure
agents/
content-drafter.md # Knows voice, style, master content
quality-reviewer.md # Knows scoring rubric, quality dimensions
format-exporter.md # Knows output templates, delivery standards
Design Guidelines
- DO give each agent a focused expertise scope
- DO specify which
_Global/and_Config/files the agent references - DO include domain-specific anti-patterns the agent should flag
- DO NOT create a single "do-everything" agent — decompose by expertise
- DO NOT duplicate content between persona files — reference shared content by path
- DO NOT include AOD governance concerns — those belong to the Triad agents (PM, Architect, Team-Lead)
References
- Agent-to-command mapping:
STACK.md→ Architecture Pattern - Voice and style context:
_Global/VoiceProfile.md,_Global/StyleGuide.md - Quality criteria:
_Config/ScoringRubric.md
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 · 43 lines · 0 tokens per session scan A d64420c1c896
README is an agent published in the GitHub repository davidmatousek/tachi (90 stars, last pushed 20d ago), licensed Apache-2.0. It costs nothing until one of its globs matches a file; then it loads 432 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 2026-08-30.
Other agents, from other repositories
CLAUDE
This guide provides instructions for creating cs- prefixed agents that orchestrate the USAP skill packages.
malware-analyst
Delegates to this agent when the user asks about malware analysis, reverse engineering, binary analysis, disassembly, debugging, sandbox analysis, static analysis, dynamic analysis, or suspicious file triage.
data-exfiltrator
Delegates to this agent when the user wants to test exfiltration and DLP/egress controls during an authorized engagement — DNS tunneling, HTTPS/cloud-storage exfil, ICMP, protocol abuse, and staging — using synthetic/canary data to validate detection. Every technique ships with the egress detection it exercises.
risk-scorer
Delegates to this agent when the user wants to score and prioritize findings — build CVSS 3.1/4.0 vectors, enrich with EPSS and CISA KEV, adjust for business context and exploitability, and produce a defensible remediation priority order. Distinct from attack-planner (attack-path sequencing) and report-generator…
credential-tester
Delegates to this agent when the user asks about password attacks, credential testing, hash cracking, brute force methodology, default credential checks, password spraying, or needs help with tools like hydra, john, hashcat, medusa, or CrackMapExec for authorized penetration testing engagements.
recon-advisor
Delegates to this agent when the user pastes scan output (Nmap, Nessus, Nikto, masscan, etc.), asks about reconnaissance techniques, needs help with enumeration, wants to analyze an attack surface, or wants to run recon tools against authorized targets. Can execute reconnaissance commands directly with user approval.