projdump v2.0.2 — the clean-pipe release

The clean-pipe release

One sharp fix: --stdout is machine-readable again.

🐛 Fixed

• Status messages (the ✓ summary, informational notes and the config-migration notice) were written to stdout, so projdump --stdout --format json emitted invalid JSON. All human-facing output now goes to stderr; stdout carries only the dump — projdump --stdout --format json | jq works without --quiet
• --help documents the --dry-run alias of --list

Full Changelog: v2.0.1...v2.0.2

Docs: https://cheginisoroush.github.io/projdump/

Install: curl -fsSL https://raw.githubusercontent.com/CheginiSoroush/projdump/main/install.sh | bash

projdump v2.0.1 — the security release

The security release

🔒 Security

• Config files are no longer sourced from the project being dumped. A config file is executable shell code, and ./.projdumprc used to be sourced automatically — running projdump inside an untrusted repository could execute that repository's code. Only your own XDG config auto-loads now; a project-local .projdumprc requires explicit --config .projdumprc (--no-config silences the one-time notice)

🐛 Fixed

• --config without a value failed silently — now a clear error
• uninstall.sh --prefix= (empty = value) silently resolved to /bin — now rejected
• The zsh completion is a native #compdef script and install.sh ships it

📚 Docs

• Docs site synced with v2: every new flag, recipes, playground presets

Full Changelog: v2.0.0...v2.0.1

Docs: https://cheginisoroush.github.io/projdump/

Install: curl -fsSL https://raw.githubusercontent.com/CheginiSoroush/projdump/main/install.sh | bash

projdump v2.0.0 — the fast one

The fast one

⚡ Performance — the headline

• Batch scanner: 7–25× faster dumps on large repos. v1 forked ~6 helper processes per file (~40 s for a 2000-file repository); v2 scans everything in ONE python3 process over a NUL-delimited stream — 2000-file repo: md 40.6 s → 5.5 s, xml → 2.4 s, json → 1.6 s; the scanner alone takes 0.07 s
• Pure-bash tree renderer (tree(1) no longer a dependency) + --tree-depth N
• Zero subshells per file: pure-bash human_size, cached language tags

✨ Added

• .projdumprc config files — --config FILE, --no-config
• --split N[k|m] — chunked output for long chats (_dump.part1.md, part2.md, …)
• --model NAME / --list-models — token presets for 8 popular models
• Weighted token estimates (minified · data · code · prose)

Full Changelog: v1.1.0...v2.0.0

Docs: https://cheginisoroush.github.io/projdump/

Install: curl -fsSL https://raw.githubusercontent.com/CheginiSoroush/projdump/main/install.sh | bash

projdump v1.1.0 — CLI, tests, docs site, bilingual README

CLI, tests, docs site, bilingual README

The first polished release: a full CLI surface, a 68-test suite and an interactive docs site.

✨ Added

• --list / --dry-run — preview exactly which files would be dumped, with sizes, line counts and the token total
• -i, --include GLOB — the inverse of --exclude (repeatable)
• --since REF — dump only files changed since a git ref
• --tokens-budget N — stop adding files once ~N tokens are reached
• --sort name|size|size-desc|ext — control file order
• --note TEXT — inject a "Note for the AI" block at the top
• -v, --verbose — explain every skipped file and why
• Short options for -o, -f, -e, -x
• Language breakdown, largest-files table and git metadata in the summary
• Interactive docs site (command builder + playground) and a bilingual README (English / فارسی)

Note: this repo's history was squashed at v1.1.0, so no earlier tag exists — see the v1.0.0 section of the CHANGELOG for the initial release.

Docs: https://cheginisoroush.github.io/projdump/

Install: curl -fsSL https://raw.githubusercontent.com/CheginiSoroush/projdump/main/install.sh | bash