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/hakimjonas/lambe/agents-mdgit clone --depth 1 https://github.com/hakimjonas/lambeWhat 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 | $0.02883 | $0.02883 |
| Opus 5 | $0.01442 | $0.01442 |
| Sonnet 5 | $0.00577 | $0.00577 |
| Haiku 4.5 | $0.00288 | $0.00288 |
Grade A, and why
lambe 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 — 247 lines — stays where its author put it; the contents beside it link to each section on GitHub.
AGENTS.md — using Lambë
Lambë (lam) is a query language for structured data. It extracts,
filters, transforms, validates, and converts JSON, YAML, TOML, HCL,
CSV, TSV, and Markdown — auto-detecting format from file extension.
This file teaches you (the agent) when to reach for lam and how to
write queries that work. The lam binary is on the user's PATH after
dart pub global activate lambe; you can invoke it from a shell tool.
When to use it
Reach for lam when the user wants to:
- Extract values from a structured file (one field, an array, a nested path).
- Filter records by a predicate.
- Transform records into a different shape.
- Aggregate numbers (sum, avg, min, max, count).
- Validate structure or values (
--assert,--schema,--explain). - Convert between formats (
--to yaml,--to csv, etc.). - Inspect unfamiliar data (
--print-shapereturns JSON Schema).
Lambë is a bounded tree transformer — every query terminates, no
recursion, no def/lambdas. Don't reach for it when the user wants:
- Binary data, images, databases, streaming.
- jq syntax specifically (use jq).
- SQL queries (use SQL).
- Programmatic processing with loops or accumulating state (write code instead).
- Recursive descent (
..),try/catch, regex,getpath/setpath, in-place mutation. See doc/non-goals.md for the full list and the lambë idiom that replaces each omission. If you hit "unknown pipe op" or a_jqIdiomHintmessage, that page is the canonical reference.
Natural language → lam query
| User says | Query |
|---|---|
| "Get the database host" | lam '.database.host' config.yaml |
| "List all user names" | lam '.users | map(.name)' data.json |
| "Filter active users over 30" | lam '.users | filter(.active && .age > 30)' data.json |
| "How many items?" | lam '.items | length' data.json |
| "Sort by price descending" | lam '.items | sort_by(.price) | reverse' data.json |
| "Group by department" | lam '.users | group_by(.dept)' data.json |
| "Total price" | lam '.items | map(.price) | sum' data.json |
| "Show the structure" | lam --print-shape data.json |
| "Check version isn't empty" | lam --assert '.version != ""' package.json |
| "Convert to YAML" | lam --to yaml '.' data.json |
| "Export as CSV" | lam --to csv '.users | map({name, age})' data.json |
| "Get all unique tags" | lam '.items | map(.tags) | flatten | unique' data.json |
| "Get the first 3 items" | lam '.items[:3]' data.json |
| "Build a summary object" | lam '{count: .items | length, total: .items | map(.price) | sum}' data.json |
| "Find containers without limits" | lam '.spec.template.spec.containers | filter(has("resources") == false) | map(.name)' deployment.yaml |
| "List Terraform resources" | lam '.resource | map(._labels)' main.tf |
| "Query CSV data" | lam '. | filter(.status != "closed") | map(.title)' issues.csv |
| "Sum a CSV numeric column" | lam '. | map(.price | to_number) | sum' orders.csv |
| "Inspect a value's type" | lam '.config | type' data.yaml |
| "List all headings in this markdown" | lam '.children | filter(.type == "heading") | map(text)' README.md |
| "What languages are in the code blocks?" | lam '.children | filter(.type == "code_block") | map(.language)' tutorial.md |
| "Run a query without input" | lam -n '[1, 2, 2, 3] | unique' |
| "Explore interactively" | lam -i data.json |
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 · 247 lines · 2,883 tokens per session scan A fdbf6345b8a1
lambe AGENTS.md is an instructions file published in the GitHub repository hakimjonas/lambe (1 stars, last pushed 2mo ago), licensed MIT. It adds 2,883 tokens to every session, about $0.0144 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
sql2json AGENTS.md
Instructions for fsistemas/sql2json, covering using sql2json with ai agents and llms, install & upgrade (for agents), or, strategy for agents and discovery — orient before querying.
sql2json CLAUDE.md
Instructions for fsistemas/sql2json, covering claude.md, read this first, commands, install dependencies and run all tests (fast, in-memory sqlite — no docker).
super-mcp-server CLAUDE.md
Instructions for berrydev-ai/super-mcp-server, covering super mcp server development guide, project overview, core architecture, code style guidelines and planning phase.
camel AGENTS.md
AGENTS.md instructions for apache/camel, covering apache camel - ai agent guidelines, project info, ai agent rules of engagement, attribution and pr volume.
wiremock AGENTS.md
AGENTS.md instructions for wiremock/wiremock, covering agent rules, running the build, messaging/websocket code and rules for these packages.
atmos CLAUDE.md
Claude Code instructions for cloudposse/atmos, covering claude.md, project overview, git worktrees (mandatory), concurrent sessions (mandatory) and hourly pr maintenance loop (recommended).