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/sandeep-alluru/agentdelta/copilot-instructionsgit clone --depth 1 https://github.com/sandeep-alluru/agentdeltaWrote 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/sandeep-alluru/agentdelta/copilot-instructions)<a href="https://agentmods.dev/instructions/sandeep-alluru/agentdelta/copilot-instructions"><img src="https://agentmods.dev/badge/instructions/sandeep-alluru/agentdelta/copilot-instructions.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 | $0.00614 | $0.00614 |
| Opus 5 | $0.00307 | $0.00307 |
| Sonnet 5 | $0.00123 | $0.00123 |
| Haiku 4.5 | $0.00061 | $0.00061 |
Grade A, and why
agentdelta copilot-instructions.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 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.
How it starts
The opening of the file, as written. The whole thing — 48 lines — stays where its author put it; the contents beside it link to each section on GitHub.
GitHub Copilot Instructions — agentdelta
agentdelta is a semantic diff engine for AI agent behavior. It records agent execution traces as JSONL, embeds each step with all-MiniLM-L6-v2, and detects the first semantic fork point between two runs.
Architecture
| Module | Purpose |
|---|---|
trace.py |
Data model: NodeType, EdgeType, TraceNode, TraceEdge, AgentTrace |
embed.py |
Sentence-transformer singleton + align_traces() sliding-window alignment |
diff.py |
diff_traces() → DiffResult, ForkPoint, StepDiff |
instrument.py |
AgentdeltaCallback (LangChain) + record() context manager |
report.py |
print_diff(), to_json(), to_markdown() formatters |
cli.py |
Click CLI: agentdelta diff and agentdelta inspect |
Key invariants
TraceNode.idis SHA-256[:16] of"{node_type}:{content}"— content-addressed, deterministicembed_trace()mutates nodes in-place; idempotentfork_threshold=0.70— cosine similarity below this → ForkPointmatch_threshold=0.85— cosine similarity above this → "match" (no change)has_regressionisTrueifffork_point is not None- Node content is truncated: LLM ≤2000 chars, tool ≤500 chars
Code style
- Python 3.10+, type-annotated, mypy strict
- Ruff rules: E, W, F, I, UP, B, S, N, SIM, RUF, PT; ignore S101 (assert in tests), N806
- No
print()in library code — userich.console.Console - All public classes and functions must have docstrings
- Tests use
pytest; CLI tests useclick.testing.CliRunner
Trace JSONL format
{"type": "trace_meta", "run_id": "v1.0"}
{"type": "node", "id": "...", "step": 1, "node_type": "start", "content": "...", "metadata": {}}
{"type": "edge", "source_step": 1, "target_step": 2, "edge_type": "sequence", "label": ""}
Adding a new output format
Add to_<format>(result: DiffResult) -> str to report.py, add to --format choices in cli.py, add tests.
Adding a new instrumentation adapter
Create src/agentdelta/instrument_<framework>.py. See instrument.py (LangChain) as reference. Export from __init__.py. Add tests.
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 · 48 lines · 614 tokens per session scan A a59f7c6c955b
agentdelta copilot-instructions.md is an instructions file published in the GitHub repository sandeep-alluru/agentdelta (0 stars, last pushed 17d ago), licensed MIT. It adds 614 tokens to every session, about $0.0031 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
langchain AGENTS.md
AGENTS.md instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
LangAlpha AGENTS.md
AGENTS.md instructions for ginlix-ai/LangAlpha, covering langalpha, common commands, architecture, backend (src/) and frontend (web/src/).
MaxKB CLAUDE.md
Claude Code instructions for 1Panel-dev/MaxKB, covering claude.md, project overview, development commands, database / static and architecture.
langchain CLAUDE.md
Claude Code instructions for langchain-ai/langchain, covering global development guidelines for the langchain monorepo, corridor security analysis, project architecture and context, monorepo structure and development tools & commands.
litellm AGENTS.md
AGENTS.md instructions for BerriAI/litellm, a project described as: The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM].
litellm GEMINI.md
Gemini CLI instructions for BerriAI/litellm, a project described as: The fastest, litest AI Gateway. Rust core with Python SDK. Call 100+ LLM APIs in OpenAI (or native) format with cost tracking, guardrails, load balancing, and logging [Bedrock, Azure, OpenAI, Anthropic, OpenAI, VertexAI, vLLM, Nvidia NIM].