mirror of https://github.com/guillaumemeyer/watermarks-remover.git synced 2026-08-22 13:11:57 +02:00

Files

T

Guillaume Meyer (The Opinionated Man)andGitHub 55d4bdc9fc feat: split skill from service, add HTTP API and Docker distribution (#60 )

* feat: split skill from service, add HTTP API and Docker distribution

The agent skill (skills/remove-ai-marks/) is now a code-free remote client:
all implementation moved to service/scripts/ and runs behind a stdlib HTTP
service (server.py) with /health, /capabilities, /inspect, /clean and a
dynamically generated OpenAPI 3.0.3 spec at /openapi.json.

- Move scripts/ and the backend Dockerfiles under service/
- server.py: JSON/base64 HTTP entrypoint with size caps, binary guard,
  atomic writes, loopback default, optional bearer auth
- Core Dockerfile (exiftool/qpdf/c2patool preinstalled) and a GHCR publish
  workflow for the core/markllm/markdiffusion images
- compose.yaml (wr-* services, harness/heavy profiles) + compose-check.sh
  to validate the running stack (exit code only)
- Fix markllm image build (tokenizers 0.22.2, CPU-only torch) and ctrlregen
  build (python:3.11 base for the 2023-era research pins)
- Fix markllm/markdiffusion harness images missing common.py at runtime

* docs: add .env.example and service configuration guide

* fix: disable chain-of-thought for openai-compatible Layer B rewrites

deepseek-v4-flash is a reasoning model: a one-line paraphrase burned 9,894
reasoning tokens (~100s) and hit the default timeout. Send
reasoning_effort=none by default for the openai-compatible backend
(--reasoning-effort / WATERMARKS_REWRITE_REASONING_EFFORT; 'off' omits the
parameter), cutting the same rewrite to ~1s / 12 tokens. Tested end-to-end
against api.deepseek.com.

* fix: sanitize client-supplied filename in HTTP service

CodeQL 'uncontrolled data in path expression' (server.py): a name like
'../../x' flowed into Path(tmpdir) / name, letting an upload escape the
request temp dir on write. Sanitize name to its basename in _decode_input
(_safe_name) and refuse any joined path whose parent is not the tmpdir at
the write sites (_tmp_path). Tests cover traversal names.

* chore: gitignore .env (contains local rewrite credentials)

* chore: deny-by-default gitignore and dockerignore; document compose env config

.gitignore and service/.dockerignore now exclude everything by default and
explicitly allow only what is publishable/needed: tracked source, docs,
tests, .github, and (for images) the service/scripts/ tree that every
Dockerfile COPYs. Root .dockerignore documents that all builds use service/
as context. README Configuration section now covers .env setup for docker
compose, host-side export for CLI runs, and the full variable table.

2026-08-14 15:42:48 -07:00

3.4 KiB

Raw Blame History

Contributing to watermarks-remover

Thanks for helping keep the skill accurate and the cleaners reliable. The project is a small Python skill (skills/remove-ai-marks/) plus tests — focused PRs land fastest.

Who can do what

Action	Who
Open issues	Anyone
Suggest a release	Anyone (use the Release suggestion issue template)
Open pull requests	Anyone (fork the repo)
Approve and merge pull requests	Maintainer only (`@guillaumemeyer`)

main is protected. A change needs a pull request, a passing CI check (test), and an approving review from the code owner before merge. Only the maintainer can give that approval. Direct pushes to main are blocked for non-admins.

To suggest a release without a code change: open a Release suggestion issue.

Prerequisites

Python 3.10+ (stdlib only for the skill scripts; optional rewrite backends use HTTP to local Ollama / OpenAI-compatible endpoints)
From the repo root: python3 -m pytest -q should pass before you open a PR
Optional for manual file checks: c2patool, exiftool (PDF)

Layout

Path	Role
`skills/remove-ai-marks/SKILL.md`	Agent skill entry (workflow, ethics) — remote client over HTTP
`skills/remove-ai-marks/references/`	Vendors, mark classes, matrix, ethics
`service/scripts/`	Layer A/B hooks + image/container cleaners + `server.py` HTTP service
`service/Dockerfile*`	Container images (core + optional backends)
`compose.yaml`	Local full-stack bring-up
`tests/`	Pytest suite and fixtures
`.github/workflows/ci.yml`	CI job `test`
`.github/workflows/release-images.yml`	GHCR image publishing on `v*` tags

Layers (what to change where)

Layer A (Unicode / format controls) — deterministic scripts under service/scripts/ (text_unicode.py, clean_text.py, inspect_text.py). Prefer tests with fixtures in tests/fixtures/.
Layer B (statistical rewrite) — guidance in SKILL.md plus optional rewrite_text.py (print-prompt default; ollama / openai-compatible). No bundled model. Keep ethics-aware.
Files (C2PA / EXIF / XMP / props) — image_meta.py (PNG/JPEG), container_meta.py (SVG/PDF/DOCX/ODT/HTML/MD), unified inspect_file.py / clean_file.py. Preserve document body / pixels; strip provenance metadata only.

Checklist for a change

Behaviour matches SKILL.md / references/removal-matrix.md when relevant
Unit tests updated or added under tests/
python3 -m pytest -q passes
Docs updated (README and/or skill references) if user-facing behaviour changes
No drive-by refactors unrelated to the fix or feature

PR expectations

Stay focused and match existing style (stdlib-first scripts, clear CLI flags)
Do not commit secrets, private user files, or large binary fixtures unless needed and redacted
Respect references/ethics.md: this tool is for content the user owns

Questions? Open an issue describing the input type (text / image / document) and which layer failed or is missing.

Community

Code of Conduct — expected behaviour in the project
Security policy — how to report vulnerabilities privately
Bug report and feature request templates

3.4 KiB Raw Blame History