Record anchored evaluation events with config snapshots and results.tsv.
npx skills add https://github.com/LinkDry/Link-Research --skill run-experiment-linkdry
Please help me install this Agent Skill. Skill: run-experiment Source: https://github.com/LinkDry/Link-Research/tree/main/skills/run-experiment Command: npx skills add https://github.com/LinkDry/Link-Research --skill run-experiment-linkdry
Records a real evaluation event as canonical evidence for anchored lines, ensuring traceability and preventing mutation of the locked claim.
Invoke the run-experiment skill to record an anchored evaluation event and generate the corresponding config snapshot and results artifact.
High-intent search queries and answers about installing and using this skill.
To record an evaluation event as canonical evidence, you identify anchored execution lines and generate an append-only artifact bundle containing the actual run configuration snapshot and results. This ensures audit-ready traceability for your research workflow.
An anchored execution line is a locked claim in your research workflow that requires traceability. You need to record its evaluation evidence when you must prevent mutation of the locked claim and prove audit-ready traceability for your experiment.
You generate a config snapshot and results.tsv by invoking the experiment recording workflow, which coordinates with project briefs and experiment memory to produce audit-ready artifacts stored in a designated workspace without mutating locked invariants.
Yes, you can use experiment memory and project briefs to preserve lineage during an evaluation run. The recording workflow coordinates with these artifacts alongside STATE.md to ensure the produced config snapshot maintains full audit-ready traceability.
To prevent mutation of locked invariants while recording experiment evidence, the workflow writes results and config snapshots to a stable, append-only location. This ensures the locked claim remains unchanged and the evidence integrity is maintained for audits.
No, the experiment recording workflow requires no external dependencies to create an audit trail. It independently identifies anchored lines and records the real evaluation event, generating the necessary config snapshot and results artifacts within your designated workspace.
LinkDryLock a write-once anchor from an experiment plan into anchor.md.
roshannandhuConvert diverse file types into a structured knowledge graph.
LinkDryArchive an experiment line while preserving lessons, artifacts, and memory.
understudy-aiCapture snapshots, clips, and motion events from RTSP/ONVIF cameras.
