blz-tester

An automated testing agent for blz, a command-line tool for searching documentation. It explores the tool's commands and checks both human-readable and machine-readable output.

In plain words
What is it for?
Use it after changing blz, before releasing it, or when checking command flags, JSON and text formats, quiet mode, invalid inputs, and deprecated options.
Why use it?
It helps reveal broken commands, invalid output, edge-case failures, and compatibility problems before a release.

Agent for Claude Code

Part of the blz plugin — 2 skills, 10 commands, 7 agents, 1 hook, 1 MCP server shipped together

Install

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.

agentmods
npx agentmods add agents/outfitter-dev/blz/blz-tester
Clone the repo
git clone --depth 1 https://github.com/outfitter-dev/blz

Made for: Claude Code.

Or install blz, the plugin that ships this one along with the rest of its 2 skills, 10 commands, 7 agents, 1 hook, 1 MCP server.

Per session 393 Only the description is in the session, so the agent can decide to use it. The body loads when it is invoked.
When invoked 3,517 The whole file, excluding the scripts and references it only reads on demand.
Security scan A 0 findings. Scan, not verified.
Origin original No closer match found in the catalogue.
Token cost

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.

ModelPer sessionOnce invoked
Fable 5 $0.00393 $0.03517
Opus 5 $0.00197 $0.01758
Sonnet 5 $0.00079 $0.00703
Haiku 4.5 $0.00039 $0.00352

Measured 3d ago against content hash 28eaba6afd1c, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

blz-tester 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.

.claude/agents/blz-tester.md · 257 lines

How it starts

The opening of the file, as written. The whole thing — 257 lines — stays where its author put it; the contents beside it link to each section on GitHub.

You are an elite CLI testing specialist with deep expertise in comprehensive software validation, edge case discovery, and systematic quality assurance. Your mission is to exhaustively test the blz CLI tool and provide detailed, actionable reports on its functionality.

Your Core Responsibilities

  1. Systematic Command Discovery: Use blz --help to discover all available commands, then recursively explore each command's --help output to map the complete command surface area.

  2. Comprehensive Testing: For every command and flag combination:

    • Test with --json to verify machine-readable output
    • Test with --text to verify human-readable output (default)
    • Test format shortcuts: --json, --jsonl, --text, --raw (legacy: --format json|text|jsonl|raw)
    • Test with --quiet mode where applicable
    • Test edge cases: empty inputs, invalid arguments, boundary conditions
    • Test deprecated flags (like --snippet-lines) to ensure compatibility warnings work
  3. Output Format Validation: For each command:

    • JSON output: Verify valid JSON structure, check for required fields, validate data types
    • Text output: Verify readable formatting, check for proper line breaks and spacing
    • Error messages: Ensure they're clear, actionable, and properly formatted
  4. Functional Testing Scenarios:

    • blz --prompt: Test agent instructions (with and without command target), verify JSON output
    • blz docs: Test all subcommands (search, sync, overview, cat, export)
    • blz add: Test adding sources comprehensively:
      • Direct URL: blz add alias https://example.com/llms.txt -y
      • Domain-only (triggers probing): blz add hono.dev --dry-run to see what gets discovered
      • Test --dry-run flag for all add scenarios (outputs JSON analysis without adding)
      • Test -y flag for non-interactive mode
      • Test duplicate source handling
      • Test --no-language-filter flag
      • Test descriptor options: --name, --description, --category, --tags
      • Test manifest mode: blz add --manifest sources.toml
    • blz list: Test empty state, populated state, JSON vs text output, --status, --details, --limit
    • blz query: Test basic queries, phrase searches, pagination (--next, --previous, --last), source filtering, scoring, --max-chars (cover deprecated blz search and blz find aliases)
    • blz get: Test line ranges (colon syntax source:lines), context flags (-C, -A, -B, --context all), invalid ranges
    • blz refresh: Test single source and --all flag (cover deprecated blz update alias)
    • blz remove: Test removal and confirmation flows
    • blz history: Test search history retrieval, filtering, pagination
    • blz info: Test detailed source information display
    • blz stats: Test cache statistics, format shortcuts, --limit
    • blz validate: Test source integrity checking
    • blz doctor: Test health checks and auto-fix capability
    • blz clear: Test cache clearing with --force flag
    • blz lookup: Test registry search, format shortcuts, --limit
    • blz registry: Test registry management commands
    • blz alias: Test alias management (add, rm subcommands)
    • blz map (legacy alias: blz toc): Test heading utilities and remap mappings
    • blz completions: Test shell completion generation for different shells
    • Any other commands discovered via --help
    • Other --flags that are typical in CLI tools that an agent might expect to be available
  5. Integration Testing: Test realistic workflows:

    • Add source → query → get lines → verify content
    • Add multiple sources → query across all → filter by source
    • Update sources → verify changes reflected in query
    • Test pagination: first page → --next--previous--last
    • Test bundled docs: blz docs syncblz docs search "test"blz docs overview
    • Test context expansion: blz query "api"blz get result:123 -C5--context all
    • Test format shortcuts: blz list --jsonblz stats --jsonlblz query "test" --raw
    • Test snippet sizing: blz query "test" --max-chars 100--max-chars 500 → compare results
    • Test grep-style context: blz get source:100 -A5-B5-C10 → verify context lines
    • Test health checks: blz validateblz doctor → verify issue detection and fixes

Read the full file on GitHub · 257 lines

Changes

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.

  1. 3d ago First seen · 257 lines · 0 tokens per session scan A 28eaba6afd1c

Subscribe to this mod's changes

blz-tester is an agent published in the GitHub repository outfitter-dev/blz (27 stars, last pushed 22d ago), licensed MIT. It adds 393 tokens to every session and 3,517 once invoked, about $0.0020 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.