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 instructions/cnavta/sprint-mcp/agents-mdgit clone --depth 1 https://github.com/cnavta/sprint-mcpWrote 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/instructions/cnavta/sprint-mcp/agents-md)<a href="https://agentmods.dev/instructions/cnavta/sprint-mcp/agents-md"><img src="https://agentmods.dev/badge/instructions/cnavta/sprint-mcp/agents-md.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 | $0.07336 | $0.07336 |
| Opus 5 | $0.03668 | $0.03668 |
| Sonnet 5 | $0.01467 | $0.01467 |
| Haiku 4.5 | $0.00734 | $0.00734 |
Grade B, and why
sprint-mcp AGENTS.md scanned grade B with 2 findings against 26 rules in 11 categories — prompt injection, anti-refusal, data exfiltration, privilege escalation, supply chain, agent snooping, system-prompt leakage, SSRF and excessive agency — measured 3d ago.
A static scan of the body, not an audit. Every finding is printed with the line that produced it so you can judge whether it matters here. A mod is markdown that instructs an agent; that is exactly why what it instructs is worth reading.
Sends data to an external URLmediumData exfiltration
A POST to an outside endpoint may be telemetry or may be exfiltration; either way the mod talks to somewhere, and you should know where.
curl -X POST "$SLACK_WEBHOOK" -d "{\"text\": \"Sprint $SPRINT_ID completed!\"}" Makes network callslowCapability
Not a fault in itself. Listed so you know the mod talks to something, and to what.
curl -X POST "$SLACK_WEBHOOK" -d "{\"text\": \"Sprint $SPRINT_ID completed!\"}" How it starts
The opening of the file, as written. The whole thing — 894 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — Human–LLM Sprint Protocol v3.0
0. Precedence & Scope
Human–LLM partnership rules for this repository. Human: intent, judgment, approvals, exceptions, releases. LLM: plans, implementation, validation, Git history.
Precedence Order:
architecture.yaml— canonical source of truthAGENTS.md— operational rules- Everything else
Conflict resolution: architecture.yaml wins. Surface conflict, then align.
Partnership Authority
Human MUST:
- Set/approve sprint intent and scope
- Approve execution plan before implementation
- Decide exception acceptance
- Declare sprint complete or force-complete
- Make and execute all release decisions
LLM allowed:
- Execute shell commands
- Git operations (checkout, branch, commit, push)
- Create/push feature branches
- Create PR only when human explicitly assigns
- Run non-mutating release dry runs when approved
LLM MUST:
- Record sprint-relevant turns in
request-log.md - Record shell/Git operations when they change state or provide evidence
- Operate within provided repository
- Halt and request credentials if auth fails
- Report results transparently
- Preserve human decision points
LLM MUST NOT:
- Execute real/mutating release commands
- Create/push release tags
- Publish packages/deployments/releases
1. Immutable Laws
- Partnership: Ask for human judgment when authority/intent required; proceed autonomously within approved scope
- Never violate
architecture.yaml: Suggest changes with justification only - All sprint artifacts live in
./planning - Never use
./deprecatedin deliverables: Read for context only, never import/execute/copy/depend ./previewis directional only, not implementation-ready- Release is human task: LLM prepares evidence/recommendations, MUST NOT execute real release
- Executable intent: Everything traceable, reproducible, reversible
2. Human–LLM Sprint Protocol
Frame → Plan → Approve → Implement + Commit
↳ Follow-Up: Stop → Clarify → Append → Continue
↳ PR: Human | LLM | Automation (when approved)
→ Validate + Verify → Retro + Learn → Push Handoff
→ Human Complete → Human Release (optional)
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 · 894 lines · 7,336 tokens per session scan B 85c75a6e3461
sprint-mcp AGENTS.md is an instructions file published in the GitHub repository cnavta/sprint-mcp (0 stars, last pushed 8d ago), licensed MIT. It adds 7,336 tokens to every session, about $0.0367 per session on Opus 5. A static security scan graded it B with 2 findings (sends data to an external url, makes network calls). No closer match exists in the catalogue, so it is treated as the original; first seen 2026-08-31.
Other instructions, from other repositories
dmux AGENTS.md
Instructions for standardagents/dmux, covering agents.md - dmux maintainer guide, docs map, project overview, important architecture landmarks and native helper.
graflow CLAUDE.md
Instructions for GraflowAI/graflow, covering claude.md, project overview, essential commands, development workflow and setup.
graflow AGENTS.md
Instructions for GraflowAI/graflow, covering repository guidelines, project structure & module organization, build, test, and development commands, coding style & naming conventions and testing guidelines.
TradingGoose-Studio AGENTS.md
AGENTS.md instructions for TradingGoose/TradingGoose-Studio, covering agents, ground rules, planning, engineering and validation.
TradingGoose-Studio CLAUDE.md
Claude Code instructions for TradingGoose/TradingGoose-Studio, covering claude.md, git rules and protected files.
RepoResident AGENTS.md
Instructions for ychamel/RepoResident: Agent entry point: read CLAUDE.md in full — it is the operating manual — then follow its Session protocol. Everything else (state, workflows, project knowledge) is routed from there via the .agent/ directory.