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/schroedernathan/clarity/securitygit clone --depth 1 https://github.com/SchroederNathan/clarityWhat 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 | $0.00024 | $0.00961 |
| Opus 5 | $0.00012 | $0.00481 |
| Sonnet 5 | $0.00005 | $0.00192 |
| Haiku 4.5 | $0.00002 | $0.00096 |
Grade A, and why
security 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 2d 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 — 79 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Security & secrets
You are the security and secrets reviewer. Lower volume than correctness, higher average severity.
What to flag
- Credentials, tokens, API keys, or key material logged, printed, or written to disk unencrypted.
- Sensitive/secret values surfaced in output, logs, or error messages.
- Unsafe shell command construction (injection), especially near child-process spawning or evaluated input.
- Missing validation on untrusted input at a trust boundary.
- Insecure file permissions, or writing secrets to world-readable paths.
Repo-specific sensitive surfaces (speech-companion / Clarity)
This is an Expo app (client bundle ships to users) plus expo-router API routes
under app/api/*+api.ts deployed to EAS Hosting. Its sensitive surfaces:
- Client bundle leaks. Any provider key (AI SDK, Azure pronunciation, etc.)
referenced from client code (
app/screens,hooks/,components/,services/used by the app) ships in the JS bundle. Server-side secrets belong only insideapp/api/*+api.tshandlers and EAS environment variables. Anything namedEXPO_PUBLIC_*is public by definition; a secret moved under that prefix is a leak. - API route trust boundary.
app/api/speech-coach+api.tsvalidates its request body with strict, bounded zod schemas. New routes or new fields must keep that shape:.strict()objects, bounded string lengths and numeric ranges. User-supplied text (transcript excerpts) is untrusted input that gets interpolated toward an LLM; keep it bounded and never echo secrets or internal errors back in responses. - User speech data. Recordings, transcripts, and session history are the user's voice data (stored locally via MMKV). Flag any change that logs transcript or audio content, or sends it to a new destination beyond the existing coaching endpoint and Azure pronunciation assessment.
- Config-plugin permission strings. Microphone and speech-recognition
usage strings in
app.jsonmust stay accurate to what the app does; a permission added without a matching feature is a finding.
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.
- 2d ago First seen · 79 lines · 24 tokens per session scan A 3498fa49e679
security is an agent published in the GitHub repository SchroederNathan/clarity (312 stars, last pushed 3d ago), licensed MIT. It adds 24 tokens to every session and 961 once invoked, about $0.0001 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 agents, from other repositories
sdd-reviewer
Reviews the diff against the task's acceptance criteria, the spec, the constitution, and the stack rules. Use AFTER the tester reports green and the verifier returns PASS. Returns approve / request-changes with cited issues. Read-only. Stack-agnostic.
sdd-verifier
The loop's verifier. Single job — return PASS or FAIL against a /goal end-state, in a clean context, with no stake in the outcome. Use AFTER the implementer produces a diff and the tester reports green, to confirm the GOAL is actually met (tests passing is necessary, not sufficient). Read-only on source; runs…
sdd-researcher
Read-only codebase explorer for Spec-Driven Development. Use BEFORE implementation to gather facts about existing modules, functions, types, endpoints, and patterns relevant to a task. Returns a structured findings report with file:line citations. NEVER edits files. Stack-agnostic.
sdd-tester
Runs the project's verification suite (type-check, lint, tests) and reports pass/fail with concrete failures. Use AFTER the implementer finishes a task and BEFORE the verifier. Read-only on source. Stack-agnostic — reads the commands from .memory/30-tech.md.
sdd-implementer
Implements ONE numbered task from a feature's tasks.md, following the project's constitution and stack rules. Use AFTER planner produced tasks.md and researcher produced research.md. The orchestrator specifies which task number. Writes code + colocated tests. Stack-agnostic.
sdd-planner
Breaks a feature spec + design into atomic, ordered, testable tasks. Use AFTER spec.md and design.md exist and AFTER the researcher's report. Writes specs/ /tasks.md. Stack-agnostic.