# Use Probity from an automated job
Use the [worked report example](evidence-test-meaning.md) to interpret test results separately from evidence conclusions. Its [retained accounting procedure](experiments/w3c-report-replay-2026-10-06/README.md) checks original files through the maintained Lab consumer.
## Find a project
Read [llms.txt](llms.txt) for the navigation index or [catalog.json](catalog.json) for structured records. Components name the tools; repository records hold their pinned README sources. Integration profiles belong to a component.
Use the stable `id` for lookups. Follow `docs_url` for the quickstart and `source.commit` to reproduce the captured source.
## Check discovery
Use the [fixed discovery prompts and receipt checker](discovery-evaluation.md) to retain a reader or agent session and review its answers against pinned public sources. Keep actual discovery attempts separate from author-operated packet checks.
## Run a local replay
To produce a new local APS/PriorSeal capture, use the [replay tools guide](replay-tools.md). Declare the caller and keep the source hashes, actual process logs and result together. The guide links one complete pinned procedure, including the copied-file route.
## Read a retained run
From a pinned Atlas checkout:
``` sh
python3 tools/check_lab.py --list
python3 tools/check_lab.py --record E6-observer-admission --json
```
The checker validates artifact hashes and measured fields before returning a record. The JSON includes results, operator roles and review state.
## Share a result
Keep the source pins, command, raw output and retained artifact together. Submit them through the [run form](https://github.com/probityai/agent-evidence-atlas/issues/new?template=evidence-run.yml), or use the [correction form](https://github.com/probityai/agent-evidence-atlas/issues/new?template=evidence-correction.yml) when a recorded claim needs fixing.
[HTML view](automation.html) | [Agent guide](llms.txt)