Truquence

Check whether supplied authorization, recovery and completion evidence agrees with a synthetic action contract before accepting an action.

Quick start

Requires Git, Python 3.10+ with venv/pip and a POSIX shell. Setup may fetch setuptools; samples run offline without an account.

git clone https://github.com/cplj/truquence-public.git
cd truquence-public
python3 -m venv .venv && . .venv/bin/activate && python3 -m pip install .
TQ_EXAMPLES="$(python3 -c 'import actioncontract; from pathlib import Path; print(Path(actioncontract.__file__).parent)')"
TQ_OUT="$(mktemp -d)"
truquence accept --contract "$TQ_EXAMPLES/contracts/synthetic.json" --results "$TQ_EXAMPLES/samples/sw.json" "$TQ_EXAMPLES/samples/ax.json" "$TQ_EXAMPLES/samples/ks.json" --out "$TQ_OUT/acceptance"

Actual sample output (JSON excerpt):

{"status":"PASS","counts":{"PASS":3,"FAIL":0,"UNKNOWN":0}}

Writes stdout JSON, $TQ_OUT/acceptance.json and .md. Hand-authored synthetic results do not prove external execution.

For AI agents

Run the commands above; replace all four input paths for your own evidence. --results order and JSON check keys: authorization=signwarden, recovery=axiocairn, completion=kinesift. Exit 0 PASS, 1 FAIL, 2 UNKNOWN, 3 output/signing error; argparse usage errors also exit 2. FAIL and UNKNOWN block acceptance. Inspect status, findings and each action's checks; never infer retry permission.

Common errors

• externally-managed-environment or command missing: activate .venv and repeat installation there.
• Missing venv/pip or old Python: provide Python 3.10+ with venv/pip.
• Missing/invalid input: correct paths in inputs/findings, then rerun.
• Output exists: retain partial files, reconcile them, then choose a fresh prefix.

Inputs and outputs · Agent reference · CSV/API

Workflows · Recorder · Modules