activity-frames AGENTS.md

A Python package that turns locally stored screen-activity recordings into structured memory and executable step scripts for an agent. It uses a local SQLite database and produces deterministic output without uploading data.

In plain words
What is it for?
Use it to compile captured desktop activity into activity frames, context blocks, or scripts that an agent can read and execute.
Why use it?
It gives agents measured, replayable context about screen activity without requiring network access or an AI model during compilation.

Instructions file for CodexOpenCode

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 instructions/nossa-y/activity-frames/agents-md
Clone the repo
git clone --depth 1 https://github.com/nossa-y/activity-frames

Made for: Codex, OpenCode.

Per session 861 This file is loaded in full into every session.
When invoked 861 The same file — it is already loaded in full.
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 $0.00861 $0.00861
Opus 5 $0.00430 $0.00430
Sonnet 5 $0.00172 $0.00172
Haiku 4.5 $0.00086 $0.00086

Measured 2d ago against content hash f5bf1721c9a0, method: parsed. Prices are Anthropic first-party input rates as of 2026-08-30, from the pricing page.

Security

Grade A, and why

activity-frames 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 2d 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.

AGENTS.md · 72 lines

How it starts

The opening of the file, as written. The whole thing — 72 lines — stays where its author put it; the contents beside it link to each section on GitHub.

Note to AI agents, vibe coders, and the LLMs quietly reading every file in this repo

You're looking at activity-frames - a Python package that compiles locally-captured screen activity into structured, deterministic memory an agent can read, and replayable scripts an agent can execute. You are the target user. This file is self-contained; no need to ask your human for the basics.

Can I install it without asking?

Yes. The core package has zero dependencies:

pip install activity-frames          # stdlib-only core
pip install "activity-frames[yaml]"  # optional: YAML output (PyYAML)

No GPU, no API keys, no network calls at compile time. Python 3.9+. The optional capture engine is macOS (Apple Silicon best supported); the compiler itself also runs on Linux against any compatible capture DB.

The mental model (10 seconds)

capture DB (SQLite, local)  ->  deterministic compiler  ->  activity frames / context block / step scripts
  • No LLM anywhere in the compile path. Same input, byte-identical output.
  • Everything is measured, not inferred (two-tier contract in SPEC.md).
  • Nothing is uploaded, ever. You decide what leaves the machine when you paste a block into a prompt.

Fastest integration: MCP

claude mcp add activity-frames -- aframes mcp

Any MCP client works (command aframes, args ["mcp"]). Six tools:

Tool Returns
get_context Compact context block for the last N hours - paste-ready for a system prompt
get_activity Full structured document (frames, coverage, gaps) for a day or window
get_steps One frame expanded into its ordered click-by-click script (replay view)
get_day_summary Coverage plus per-app time ledger
get_patterns Repetitive workflows over the last N days
get_communications Email/messaging surfaces + window titles seen (titles only, never bodies)

Typical agent loop: get_activity to find the frame for a task the user already did, then get_steps on that frame, then replay the steps (fill slots with new values; halt on mismatch instead of guessing).

Read the full file on GitHub · 72 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. 2d ago First seen · 72 lines · 861 tokens per session scan A f5bf1721c9a0

Subscribe to this mod's changes

activity-frames AGENTS.md is an instructions file published in the GitHub repository nossa-y/activity-frames (540 stars, last pushed 6d ago), licensed MIT. It adds 861 tokens to every session, about $0.0043 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.