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/drbaher/compare-cli/agents-mdgit clone --depth 1 https://github.com/DrBaher/compare-cliWhat 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 | $0.03157 | $0.03157 |
| Opus 5 | $0.01579 | $0.01579 |
| Sonnet 5 | $0.00631 | $0.00631 |
| Haiku 4.5 | $0.00316 | $0.00316 |
Grade A, and why
compare-cli AGENTS.md 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 yesterday.
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 — 327 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md
This file describes the stable surface of compare-cli for LLM agents
and downstream tooling. The exit codes and --json shape documented here
are stable across v1.x minor releases.
Output contract
compare-cli is a deterministic pre-signature gate: given two contract
versions (BASE = what was agreed, CANDIDATE = what's being put forward),
it classifies the drift between them and reports it via the exit code.
stdout is the report. Human-readable when stdout is a TTY (with ANSI
color), structured JSON when --json is set, plain text otherwise.
stderr is the side channel. Warnings, the --why block, and error
messages go to stderr. --silent / -q suppresses all stderr output;
exit codes are unchanged. Argument-parse errors always print on the real
stderr regardless of --silent.
Exit codes
The exit code is the verdict — these are drift severities, not just error/success, so branch on them (don't assume the suite's common scheme):
| Exit | exit_class |
Meaning |
|---|---|---|
0 |
clean |
No drift detected. Every clause matches. Safe to sign. |
1 |
n/a | I/O error — input not found, unreadable, malformed .docx/.pdf. |
2 |
substantive |
Substantive drift, or --strict/--strict-cosmetic was tripped. |
3 |
cosmetic / typographic |
Cosmetic-only or typographic-only drift. Informational. |
4 |
moved |
Clause(s) moved but content identical. |
--json output shape (stable across v1.x)
{
"ok": true,
"exit_class": "clean",
"exit_code": 0,
"base": {
"path": "negotiated.docx",
"format": "docx",
"lossiness": "none",
"clauses_total": 12,
"track_changes": []
},
"candidate": {
"path": "ready-to-sign.pdf",
"format": "pdf",
"lossiness": "extracted",
"clauses_total": 12,
"track_changes": []
},
"summary": {
"clauses_total": 12,
"clauses_changed": 0,
"clauses_moved": 0,
"clauses_added": 0,
"clauses_removed": 0,
"differences": {
"cosmetic": 0,
"typographic": 0,
"substantive": 0,
"added": 0,
"removed": 0,
"moved": 0
},
"suppressed_by_filter": 0
},
"differences": [],
"warnings": []
}
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.
- yesterday First seen · 327 lines · 3,157 tokens per session scan A b5addc944b96
compare-cli AGENTS.md is an instructions file published in the GitHub repository DrBaher/compare-cli (1 stars, last pushed 29d ago), licensed MIT. It adds 3,157 tokens to every session, about $0.0158 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 2026-08-31.
Other instructions, from other repositories
l4-ide AGENTS.md
AGENTS.md instructions for smucclaw/l4-ide, covering agents.md, about l4, repository structure, haskell (cabal) and typescript (npm workspaces + turborepo).
korean-law-mcp CLAUDE.md
Instructions for chrisryugj/korean-law-mcp, covering claude.md, structure, commands, cli usage (v2.0) and 자연어 한 줄로 법령 조회.
blazediff AGENTS.md
Instructions for teimurjan/blazediff, covering blazediff agent rules, hard rules, quick commands and pointers.
critique AGENTS.md
Instructions for remorses/critique, covering opentui, using unreleased opentui versions, bun, react and rules.
traverse AGENTS.md
Instructions for traverse-framework/traverse, covering traverse — agent coordination, agent coordination, 1. check for claude code claim, 2. check for claude code branch and 3. claim the ticket (only if pre-flight passes).
open-agreements AGENTS.md
Instructions for open-agreements/open-agreements, covering repository guide and local mcp servers.