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/quickcall-dev/blackbox/agents-mdgit clone --depth 1 https://github.com/quickcall-dev/blackboxWhat 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 | $0.00641 | $0.00641 |
| Opus 5 | $0.00320 | $0.00320 |
| Sonnet 5 | $0.00128 | $0.00128 |
| Haiku 4.5 | $0.00064 | $0.00064 |
Grade A, and why
blackbox 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 — 57 lines — stays where its author put it; the contents beside it link to each section on GitHub.
Blackbox – Agent Guide
What this repo is
Blackbox is the analysis engine inside QuickCall. It takes raw AI coding session traces (JSONL), runs a multi-stage LLM pipeline to find root causes, recurring failures, and anti-patterns, then feeds that back so future agent sessions improve.
How to run
uv sync --extra dev
uv run uvicorn src.main:app --host 0.0.0.0 --port 8000
How to test
uv run pytest # all tests
uv run pytest tests/test_api.py -x # just API tests, stop on first failure
Key files to know
| File | What it does |
|---|---|
src/main.py |
FastAPI app, routes, file upload, source detection, normalization dispatch |
src/config.py |
Settings from env/.env |
src/models.py |
Pydantic response models |
src/pipeline/orchestrator.py |
Pipeline class — runs P0–P6 stages in order |
src/pipeline/annotator.py |
Annotates raw messages with tool calls, edits, thinking blocks |
src/pipeline/context_builder.py |
Builds context windows around trigger turns |
src/pipeline/dedup.py |
Deduplicates findings by similarity threshold |
src/classify/runner.py |
Classification runner — labels user messages via LLM |
src/classify/prompts.py |
Classification system prompt + JSON schema |
src/rca/prompts.py |
All analysis prompts and JSON schemas (RCA, behavior, cluster, convention) |
src/normalizer/unified.py |
NormalizedMessage — the universal format all sources map to |
src/llm/client.py |
Async LLM client (OpenAI-compatible API) |
src/storage/run_store.py |
In-memory store for run state and stage outputs |
Architecture
See docs/architecture.md for full Mermaid diagram, file structure, data flow, and pipeline stage details.
How to add a new source format
- Add transform in
src/normalizer/<source>_transform.py - Add
normalize_<source>()function returninglist[NormalizedMessage] - Import and wire into
_normalize_file()insrc/main.py - Add detection heuristics in
_detect_source()insrc/main.py
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 · 57 lines · 641 tokens per session scan A 06c6bb39f24c
blackbox AGENTS.md is an instructions file published in the GitHub repository quickcall-dev/blackbox (3 stars, last pushed 2mo ago), licensed Apache-2.0. It adds 641 tokens to every session, about $0.0032 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
full-stack-ai-agent-template CLAUDE.md
Instructions for vstorm-co/full-stack-ai-agent-template, covering claude.md, project overview, commands, install dependencies and run tests.
full-stack-ai-agent-template AGENTS.md
Instructions for vstorm-co/full-stack-ai-agent-template, covering agents.md, project overview, commands, cli and architecture.
maggy CLAUDE.md
Claude Code instructions for alinaqi/maggy, covering global claude code settings, general preferences, code style, git workflow and tools & frameworks.
taskdog CLAUDE.md
Instructions for Kohei-Wada/taskdog, covering claude.md, project overview, monorepo structure, data storage & configuration and development commands.
python-samples-fastapi-restful CLAUDE.md
Instructions for nanotaboada/python-samples-fastapi-restful, covering claude.md, claude code, tech stack, structure and coding guidelines.
txtify CLAUDE.md
Instructions for lkmeta/txtify, covering txtify — agent guide, architecture (read this before touching job flow), commands, invariants & known traps and conventions.