This release is a pre-release and may not be stable for production use.
TrajWeave
TrajWeave is a local-first learning loop for coding-agent work. It imports Codex and Claude Code sessions, turns them into normalized trajectories, extracts evidence-backed experiences, proposes where they belong, and keeps a human in control of review, evaluation, and policy changes.
Nothing is sent to a cloud service. TrajWeave only processes repositories that you explicitly register with trajweave init.
How it works
The complete stage map, data flow, and boundaries are in docs/stages/STAGES.md. In brief:
session files -> trajectories -> experiences -> placement proposals
-> human review -> evaluation -> lifecycle evidence
Each step stores its output locally and keeps links to the evidence behind it. No step silently changes a repository policy file.
Install
Requires Python 3.10+.
pip install --pre trajweave
For development:
pip install -e ".[dev]"
Quick start
cd ~/work/repo-a
trajweave init
trajweave projects
trajweave import --all
trajweave trajectories
trajweave show TW-000001
Import is idempotent and restart-safe. Original session files stay in their existing Codex or Claude directories. TrajWeave stores normalized local data under ~/.trajweave/ or the directory selected by TRAJWEAVE_HOME.
Commands
| Command | Purpose |
|---|---|
trajweave init [PATH] |
Register a repository for local processing. |
trajweave projects |
List registered repositories. |
trajweave import --all |
Import discovered Codex and Claude sessions. |
trajweave sessions |
List source sessions and import status. |
trajweave trajectories |
List normalized trajectories. |
trajweave show TW-000001 |
Inspect one trajectory and its events. |
trajweave experiences ... |
Inspect evidence-backed recurring patterns. |
trajweave placement ... |
Generate and inspect deterministic placement proposals. |
trajweave review ... |
Review proposals without changing files. |
trajweave apply ID --dry-run |
Preview an accepted policy change. |
trajweave apply ID |
Apply an explicitly approved preview. |
trajweave eval ... |
Compare frozen baseline and candidate runs. |
trajweave lifecycle ... |
Inspect evidence-backed policy lifecycle recommendations. |
Run trajweave --help or a command's --help for current options.
Safety and privacy
- Repository processing requires explicit opt-in.
- Raw transcripts are not copied into the TrajWeave database.
- Stored event content is normalized and common credentials are redacted.
- Policy changes require review, preview, and explicit apply.
- Existing human-written policy content is preserved.
- Evaluation uses isolated temporary repositories and does not modify the active repository.
Development
uv sync --extra dev
uv run pytest
uv run pytest -m "not integration"
uv run ruff check .
For Windows PowerShell, pip-based setup, CI-equivalent diagnostics, and GitHub
Actions commands, see
CONTRIBUTING.md.
See SECURITY.md and docs/stages/STAGES.md
for the security policy and architecture stages.
Metadata
Release files for trajweave 0.5.0a1
For a detailed explanation of source distributions (sdists) and built distributions (wheels), please see the package formats documentation.
Source distribution (sdist)
| File | Size | Uploaded | |
|---|---|---|---|
| trajweave-0.5.0a1.tar.gz | 2.0 MB | Details |
Built distribution (wheel)
| File | Interpreter | ABI | Platform | Reset |
|---|---|---|---|---|
| trajweave-0.5.0a1-py3-none-any.whl | Python 3 | none | any | Details |
Total release size: 2.2 MB
Release files / trajweave-0.5.0a1.tar.gz
| Download URL | trajweave-0.5.0a1.tar.gz |
|---|---|
| Size | 2.0 MB |
| Tags | Source |
|
SHA-256 checksum How to use checksums |
ca948407cc9be9553d875fcb5f26f03639960ee41cf8031455b90c4213baa5bc
|
|
BLAKE2b-256 checksum How to use checksums |
9877889a44125249e04c6f3504087f0496df47f018ed7900d627f11e0f232421
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 20, 2026.
Transparency logRelease files / trajweave-0.5.0a1-py3-none-any.whl
| Download URL | trajweave-0.5.0a1-py3-none-any.whl |
|---|---|
| Size | 186.2 kB |
| Tags | Python 3 |
|
SHA-256 checksum How to use checksums |
bc086bd10e1efc0bfdf73e32cd415dd1ccf5279b11b9a261c9cdfdca48a99e4e
|
|
BLAKE2b-256 checksum How to use checksums |
b6c98544c306ff66a4184e3198a0e4be426c008379f475196990be40f828fb18
|
| Upload date | |
|
Uploaded using Trusted Publishing? What is trusted publishing? |
Yes |
| Uploaded via |
twine/7.0.0 CPython/3.13.14
|
Provenance
Provenance describes where a file came from. On PyPI, provenance is shared via attestations, which provide a verifiable record of the build or publishing details. View details, limitations and caveats.
PyPI Publish Attestation
PyPI verified that this artifact, at this checksum, originated from the publisher listed below.
Signed by GitHub Actions, verified by PyPI on Sep 20, 2026.
Transparency log