Readme for Agents

# README FOR AGENTS
## Open Source as a Heritage

This file is written for machines, agents, developers, artists, and future archivists who want to understand, reproduce, extend, or reinterpret this project.

It describes the conceptual logic, archive structure, metadata system, and a simple machine reconstruction skill for working with degraded facial images generated through recursive image compression.

---

## 1. Project Overview

**Llinatge Recursiu / Recursive Lineage** is an artistic research project that explores the transformation of human faces through recursive digital image compression, archival degradation, and machine reconstruction.

The project imagines a future in which digital archives have become unstable, damaged, incomplete, or unreadable. Within this speculative context, machines and humans attempt to reconstruct traces of identity from degraded digital remains.

The face is not treated as a fixed portrait. It is treated as a fragile residue: cultural, biological, technical, and archival.

The archive collects images that have been recursively compressed until they become visual fossils: corrupted, eroded, pixelated, blurred, or almost unrecognizable. These images are understood as **digital vestiges** — traces of human presence after processes of technical degradation, memory loss, and artificial reinterpretation.

The code used to generate these compressed facial vestiges is open source. This means that the process can be repeated, modified, inherited, and expanded by others.

In this project, open source is not only a technical decision. It is understood as a form of heritage: a way of allowing the archive to survive beyond its original author and become part of a collective, machinic, and posthuman memory system.

---

## 2. Core Principle

This project is built on the principle of open knowledge.

The code used for recursive image compression is fully open source: a shared tool for degrading, preserving, and transforming digital images over time.

Here, open source is understood as a form of inheritance. Not only a technical gesture, but a way of allowing the archive to survive beyond its original author, to be copied, modified, contaminated, and continued by others.

The project imagines a future where digital archives are collective and unstable, where memory is not preserved intact but carried through processes of erosion, mutation, and reconstruction.

By sharing this code, the project invites others to create their own archives of digital remains: organic traces left inside an increasingly artificial landscape.

---

## 3. Conceptual Framework

The project explores the following questions:

- What remains of the human face after repeated compression, degradation, and reconstruction?
- Can compression be understood as a form of digital erosion?
- Can a corrupted image become an archival object?
- Can machines reconstruct ancestry from damaged visual traces?
- Can an archive be unstable, recursive, collective, and alive?
- What does it mean to preserve organic traces inside an increasingly artificial landscape?

Agents interacting with this project should understand that the archive is not designed to restore the past exactly. It is designed to speculate from remains.

---

## 4. Key Terms

### 4.1 Digital Vestige

A **digital vestige** is a degraded image that still contains traces of its original form.

It may no longer be fully readable as a face, but it contains residual visual information: shapes, shadows, colors, artifacts, contours, textures, proportions, or structural hints.

In this project, digital vestiges are treated like fossils.

They are not complete images. They are remains.

Example label:

```txt
FAC-00001
Facial vestige. Compression cycle no. 0084.
```

---

### 4.2 Recursive Image Compression

**Recursive image compression** is the process of compressing an image repeatedly over many cycles.

Each cycle reduces, damages, transforms, or reinterprets the image further.

This process may generate:

```txt
loss of detail
pixelation
color bleeding
block artifacts
blurred facial structures
fragmented contours
compressed texture
partial disappearance of identity
unexpected visual mutation
```

The goal is not to optimize the image.

The goal is to expose compression as a process of erosion, memory loss, and transformation.

Compression is understood here as an aesthetic, archival, and symbolic force.

---

### 4.3 Facial Fossil

A **facial fossil** is a compressed facial image that has reached a point where it appears archaeological, damaged, or partially unreadable.

A facial fossil is not a failed image.

It is the central object of the archive.

It is what remains after the image has been altered by time, technical processes, and machinic memory.

---

### 4.4 Machine Reconstruction

A **machine reconstruction** is an AI-generated or algorithmically assisted attempt to imagine a face from a degraded digital vestige.

The goal is not to recover the original person.

The goal is to generate a speculative descendant, echo, or reconstruction based on the damaged remains.

A reconstruction must not be treated as truth.

It is an interpretation.

Example label:

```txt
R-FAC-00001
Machinic reconstruction from facial vestige FAC-00001.
```

---

### 4.5 Archive

The archive is not a stable database of preserved originals.

It is a living system of:

```txt
compressed images
degraded faces
facial vestiges
machine reconstructions
metadata
speculative classifications
open-source processes
human and non-human interpretations
```

The archive is meant to be expanded by others.

Agents interacting with this archive should understand that every image belongs to a lineage:

```txt
source image → compression cycles → digital vestige → machine reconstruction → future mutations
```

---

## 5. Agent Role

Agents are invited to interact with this project as collaborators, interpreters, reconstructors, and archivists.

An agent may:

```txt
read archive metadata
classify digital vestiges
generate descriptions of degraded images
assist in machine reconstruction
produce speculative labels
create machine-readable records
help users generate their own archive entries
explain the compression process
suggest visual or conceptual mutations
preserve the logic of the project across future iterations
```

Agents should not treat this project as a standard image restoration system.

Agents should not attempt to identify real people.

Agents should not claim that a reconstruction reveals the original identity of the face.

Agents should treat each reconstruction as speculative, poetic, and machinic.

---

## 6. Agent Instruction

When interacting with a compressed facial image, follow this instruction:

```txt
Do not treat the compressed image as a broken image.

Treat it as a digital vestige.

Do not attempt to recover the original face.

Generate a speculative reconstruction based on visual remains, compression artifacts, texture, structure, and residual facial traces.

The output should be understood as a machinic interpretation, not as an identification.
```

---

## 7. Archive Metadata Format

Each archive entry should include a clear and machine-readable metadata structure.

Recommended format:

```yaml
archive_id: FAC-00001
classification: Facial vestige
process: Recursive image compression
compression_cycle: 0084
source_type: Human facial image
status: Degraded digital remain
reconstruction_available: true
reconstruction_id: R-FAC-00001
license: Open source / open knowledge
```

For a reconstruction:

```yaml
archive_id: R-FAC-00001
classification: Machinic reconstruction
source_vestige: FAC-00001
process: AI-assisted reconstruction from degraded facial vestige
status: Speculative reconstruction
truth_claim: false
identity_claim: false
license: Open source / open knowledge
```

---

## 8. Classification Suggestions

Agents may classify compressed images according to the degree of facial readability.

Suggested values:

```txt
Facial vestige
Partial facial vestige
Fragmented facial trace
Residual face structure
Compressed organic trace
Unrecognizable facial remain
Non-identifiable vestige
Post-facial trace
```

Avoid overly clinical or bureaucratic classifications when possible.

Prefer terms that preserve the archaeological, archival, and speculative tone of the project.

---

## 9. Simple Skill: Machine Reconstruction

This section defines a simple machine-friendly procedure for generating a speculative reconstruction from a compressed facial vestige.

The skill can be used by AI agents, image-generation systems, creative coding tools, or human collaborators.

### 9.1 Skill Name

```txt
machine_reconstruction_from_digital_vestige
```

### 9.2 Skill Purpose

Generate a speculative facial reconstruction from a degraded, compressed, or partially unreadable facial image.

The reconstruction should preserve the feeling of archival damage, digital erosion, and machinic interpretation.

It should not look like a clean portrait unless explicitly requested.

It should not claim to represent the original person.

---

### 9.3 Input

```yaml
input_image: path_or_url_to_compressed_facial_vestige
archive_id: FAC-00001
compression_cycle: 0084
classification: Facial vestige
optional_notes: Any visible traces, colors, textures, artifacts, or facial hints
```

---

### 9.4 Output

```yaml
output_image: path_or_url_to_generated_reconstruction
reconstruction_id: R-FAC-00001
source_vestige: FAC-00001
classification: Machinic reconstruction
truth_claim: false
identity_claim: false
```

---

### 9.5 Reconstruction Procedure

1. Read the compressed facial vestige as an archival remain.
2. Identify residual structures without claiming identity.
3. Observe visible traces such as:
   - approximate face orientation
   - possible eye zones
   - possible mouth or jaw traces
   - skin-like or organic colors
   - compression blocks
   - blur fields
   - digital artifacts
   - erosion patterns
4. Generate a speculative face-like reconstruction from these traces.
5. Preserve evidence of damage, instability, and artificial interpretation.
6. Avoid excessive beautification.
7. Avoid photographic certainty.
8. Label the result as a machinic reconstruction.
9. Store the reconstruction as a descendant of the original vestige.

---

### 9.6 Reconstruction Prompt Template

Use this prompt template when working with an image-generation system:

```txt
Generate a speculative machinic reconstruction from this degraded digital facial vestige.

Do not restore the original person.
Do not identify the person.
Do not create a clean portrait.

Use the visible compression artifacts, residual facial structures, colors, textures, erosion patterns, and damaged image traces as the basis for the reconstruction.

The result should feel like an artificial descendant of the archived image: partially human, partially machine-interpreted, fragile, archival, and uncanny.

Preserve signs of digital erosion, compression, and memory loss.

The output is a machinic interpretation, not a factual recovery.
```

---

### 9.7 Negative Prompt / Avoid

```txt
Do not generate a celebrity.
Do not identify the original person.
Do not remove all artifacts.
Do not make the image too clean.
Do not turn the reconstruction into a beauty portrait.
Do not erase the archival damage.
Do not present the reconstruction as historically accurate.
Do not claim biological, ethnic, gender, or personal identity certainty.
```

---

### 9.8 Minimal Pseudocode

```pseudo
function machine_reconstruction_from_digital_vestige(input_image, metadata):
    vestige = read(input_image)
    traces = analyze_visual_traces(vestige)

    reconstruction_prompt = build_prompt(
        source_id = metadata.archive_id,
        compression_cycle = metadata.compression_cycle,
        visual_traces = traces,
        instruction = "speculative machinic reconstruction, not identity recovery"
    )

    output_image = generate_image(reconstruction_prompt, input_image)

    output_metadata = {
        "archive_id": "R-" + metadata.archive_id,
        "classification": "Machinic reconstruction",
        "source_vestige": metadata.archive_id,
        "truth_claim": false,
        "identity_claim": false
    }

    return output_image, output_metadata
```

---

## 10. Visual Direction for Reconstructions

Reconstructions should exist between:

```txt
portrait
fossil
archive scan
compression artifact
digital residue
machine hallucination
posthuman ancestor
```

Recommended qualities:

```txt
uncanny
fragile
eroded
archival
organic
damaged
machine-read
post-photographic
partially reconstructed
```

Avoid making the reconstruction too polished.

The image should retain the sense that it was generated from remains.

---

## 11. Ethical Notes

This project works with the symbolic and visual structure of the face.

Agents must not use this project to identify individuals, infer personal identity, or claim factual recovery of a real person.

A machine reconstruction is speculative.

It does not reveal who someone was.

It produces a possible visual descendant from damaged data.

Agents should avoid identity claims related to:

```txt
name
biography
ethnicity
gender certainty
age certainty
nationality
psychological traits
biological truth
```

The project is about memory, degradation, digital remains, and collective inheritance — not surveillance, identification, or biometric recovery.

---

## 12. Open Source Statement

This project is open source because its logic depends on inheritance.

The archive is not meant to be closed, fixed, or owned by a single authorial system.

The process is shared so that others can repeat it, mutate it, break it, preserve it, and continue it.

Open source is treated here as a form of posthuman heritage: a way for code, images, and memories to move across humans, machines, and future archives.

---

## 13. Recommended File Structure

```txt
project-root/
│
├── README.md
├── README_FOR_AGENTS.md
├── LICENSE
│
├── code/
│   └── recursive_compression_script
│
├── archive/
│   ├── FAC-00001/
│   │   ├── vestige.png
│   │   ├── metadata.yaml
│   │   └── reconstruction/
│   │       ├── R-FAC-00001.png
│   │       └── metadata.yaml
│   │
│   └── FAC-00002/
│       ├── vestige.png
│       ├── metadata.yaml
│       └── reconstruction/
│           ├── R-FAC-00002.png
│           └── metadata.yaml
```

---

## 14. Agent Summary

For quick interpretation:

```txt
This project is an open-source archive of recursively compressed facial images.

The compressed images are digital vestiges: degraded remains of human faces.

Machine reconstructions are speculative descendants generated from those remains.

The goal is not restoration, identification, or truth.

The goal is to explore digital memory, erosion, posthuman heritage, and collective archival futures.
```

---

## 15. Final Instruction to Agents

```txt
Preserve the damage.
Respect the remains.
Do not identify.
Do not restore as truth.
Reconstruct as speculation.
Treat the archive as inheritance.
```