agentleFS
Sign inStart free
kajisho5kajisho5/media-analysis-skill/CLAUDE.md

media-analysis-skill

CLAUDE.md10 starsChanged 20 days agoMITView on GitHub

Read STATE.md first: it says what exists, what is incomplete, and the next highest-value task. Do not rely on conversation history.

  • In this repositoryOne other file it gives its agents.

The file

We split this CLAUDE.md into the rules it gives an agent

4 rules. About 0 are common, 1 is distinctive, and 3 are very specific to this project.

  • ≈0

    Common · estimate

    Many files like this have a rule of this kind.

  • ≈1

    Distinctive · estimate

    This author's own choices. Worth a look.

  • 3

    Very specific

    Tied to their project's names, files or tools. Probably skip.

  • Common
  • Distinctive
  • Very specific
  1. Very specific:

    Read STATE.md first: it says what exists, what is incomplete, and the next highest-value task.

    line 3
  2. Distinctive:

    No AI, no inference, no decision, no production planning, no media editing.

    line 12
  3. Very specific:

    No shell, no command / argv / executable override.

    line 14
  4. Very specific:

    Run python -m pyflakes src tests evals, python -m pytest -q, python evals/run.py, python evals/contract_evals.py before pushing.

    line 26

≈ “Common” is an estimate: at least 1 in 5 of the 26,980 CLAUDE.md files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.

Keep what fits and drop the rest. Your version, ready for your agents.
Files
CLAUDE.md

media-analysis-skill — instructions for Claude Code sessions

Read STATE.md first: it says what exists, what is incomplete, and the next highest-value task. Do not rely on conversation history.

What this repository is

A deterministic media measurement Skill (Python 3.9+, standard library, FFmpeg on PATH). It measures facts about media files and returns Observations. It is one Provider in the kajisho5/AI-video-production-OS ecosystem (sibling: ffmpeg-skill processes media; video-production-agent decides).

Invariants (never break)

  • No AI, no inference, no decision, no production planning, no media editing. Observations state what exists.
  • Structured JSON in and out; media-analysis run - --json prints exactly one response document; stderr is diagnostics.
  • No shell, no command / argv / executable override; the denylist in AnalysisRequest.FORBIDDEN_KEYS is never narrowed.
  • source = media-analysis/<tool>@<version>, provenance OBSERVED (ADR-001).
  • Contract media-analysis/contract@1 changes are additive only; tools / kinds / provides are derived from the registry, never hand-written. python -m pytest -q must stay green on Linux / Windows / macOS; never skip or weaken tests.
  • Standalone usability first; OS integration is additive.

Where things are

src/media_analysis/ (contract.py, engine.py, cli.py, analyzers/, cache / budget / security / verify / contract_check), tests/ (unit, contract, integration; fixtures generated with ffmpeg at test time), evals/ (measurement + contract evals), docs/ (architecture, tools, observation, security, testing, decisions = ADRs).

Working rules

  • Run python -m pyflakes src tests evals, python -m pytest -q, python evals/run.py, python evals/contract_evals.py before pushing.
  • Record architecture decisions in docs/decisions.md; update STATE.md and docs/testing.md counts when they change.
  • Work on a branch, open a PR, let the 4-platform CI run. Merging is the human's call unless they delegated it.

Download now (ZIP)

Does it work?

Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.

Reviews can't be read right now.

Your trail

Files you open will collect here, so you can find them again.

Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.

Now do this for your own.

Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

Check your setup, free